CN104504032A - Method and equipment for providing service upon user emotion tendencies - Google Patents

Method and equipment for providing service upon user emotion tendencies Download PDF

Info

Publication number
CN104504032A
CN104504032A CN201410773678.3A CN201410773678A CN104504032A CN 104504032 A CN104504032 A CN 104504032A CN 201410773678 A CN201410773678 A CN 201410773678A CN 104504032 A CN104504032 A CN 104504032A
Authority
CN
China
Prior art keywords
text
assessed
user
time
probability distribution
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
Application number
CN201410773678.3A
Other languages
Chinese (zh)
Other versions
CN104504032B (en
Inventor
于魁飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Zhigu Ruituo Technology Services Co Ltd
Original Assignee
Beijing Zhigu Ruituo Technology Services Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Zhigu Ruituo Technology Services Co Ltd filed Critical Beijing Zhigu Ruituo Technology Services Co Ltd
Priority to CN201410773678.3A priority Critical patent/CN104504032B/en
Publication of CN104504032A publication Critical patent/CN104504032A/en
Application granted granted Critical
Publication of CN104504032B publication Critical patent/CN104504032B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Machine Translation (AREA)

Abstract

The embodiment of the invention discloses a method for providing service upon user emotion tendencies. The method is characterized by comprising the following steps of determining a to-be-evaluated text, wherein the to-be-evaluated text is at least one text read by a user; according to a model that user emotion evolves along with time, determining probability distribution that the to-be-evaluated text enables the user to produce every user emotion tendency; according to the probability distribution that the to-be-evaluated text enables the user to produce every user emotion tendency, providing the service for the user. The invention also discloses equipment for providing the service upon the user emotion tendencies. By adopting the method and the equipment for providing the service upon the user emotion tendencies, disclosed by the application, deep analysis can be performed on the user emotion tendencies, accordingly the pointed service is provided, and the defect of the prior art is remedied.

Description

The method and apparatus of service is provided based on user feeling tendency
Technical field
The application relates to data mining technology field, particularly relates to a kind of method and apparatus providing service based on user feeling tendency.
Background technology
In recent years, along with the development of internet correlation technique, various content pushes the Main Means becoming application provider's adding users frequency of utilization gradually.Current have much based on the content propelling movement mode of user behavior.Such as, the commodity browsed according to user, push similar commercial promotions information to user; According to the current location of user, push neighbouring information on services etc. to user.
Current, many portal websites start to allow user to share their mood about certain news.By analyzing user to the Sentiment orientation of certain news, can know that user sees that what news can be glad, seeing that what news can indignation, see that what news can be dejected etc.
Which type of Sentiment orientation user produces after reading text, reflects the point of interest place of user to a great extent.But existing content pushes mode certain limitation, cannot know the Sentiment orientation differentiation in time of user, more cannot carry out the propelling movement of content based on this.
Summary of the invention
The object of the application is: provide a kind of method and apparatus providing service based on user feeling tendency.
According to an aspect of at least one embodiment of the application, provide a kind of method that service is provided based on user feeling tendency, comprising:
Determine text to be assessed, wherein, described text to be assessed is at least one text that a user read;
According to the model that a user feeling develops in time, determine that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation;
Described user is made to produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service.
According to another aspect of at least one embodiment of the application, a kind of equipment providing service based on user feeling tendency is provided, comprises:
One assessment text determining device, for determining text to be assessed, wherein, described text to be assessed is at least one text that a user read;
One emotion probability distribution determining device, for the model developed in time according to a user feeling, determines that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation;
One service providing apparatus, for making described user produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service.
Adopt the method and apparatus that service is provided based on user feeling tendency described in the application, depth analysis can be carried out to user's feelings tendency, and provide accordingly and serve targetedly, compensate for the defect of prior art.
Accompanying drawing explanation
Fig. 1 is the method flow schematic diagram setting up emotion model that an embodiment of the application provides;
Fig. 2 is the interactive interface schematic diagram of user feedback Sentiment orientation in an embodiment of the application;
Fig. 3 a is the method flow schematic diagram setting up emotion model that another embodiment of the application provides;
Fig. 3 b is the method flow schematic diagram setting up emotion model that another embodiment of the application provides;
Fig. 4 is the method flow schematic diagram of the prediction user feeling tendency that the application's embodiment provides;
Fig. 5 is the method flow schematic diagram of the monitoring public sentiment that the application's embodiment provides;
Fig. 6 is the method flow schematic diagram providing service based on user feeling tendency that the application's embodiment provides;
Fig. 7 is the apparatus structure schematic diagram setting up emotion model that the application's embodiment provides;
Fig. 8 is the apparatus structure schematic diagram setting up emotion model that another embodiment of the application provides;
Fig. 9 is the apparatus structure schematic diagram setting up emotion model that another embodiment of the application provides;
Figure 10 is the apparatus structure schematic diagram setting up emotion model that another embodiment of the application provides;
Figure 11 is the first output sub-module 744 or the second output sub-module 748 structural representation that the application's embodiment provides;
Figure 12 is the device structure schematic diagram of the prediction user feeling tendency that the application's embodiment provides;
Figure 13 is the device structure schematic diagram of the prediction user feeling tendency that another embodiment of the application provides;
Figure 14 is the device structure schematic diagram of the prediction user feeling tendency that another embodiment of the application provides;
Figure 15 is the device structure schematic diagram of the prediction user feeling tendency that another embodiment of the application provides;
Figure 16 is the device structure schematic diagram of the monitoring public sentiment that the application's embodiment provides;
Figure 17 is the device structure schematic diagram of the monitoring public sentiment that another embodiment of the application provides;
Figure 18 is the device structure schematic diagram of the monitoring public sentiment that another embodiment of the application provides;
Figure 19 is the device structure schematic diagram of the monitoring public sentiment that another embodiment of the application provides;
Figure 20 is the device structure schematic diagram of the monitoring public sentiment that another embodiment of the application provides;
Figure 21 is the device structure schematic diagram providing service based on user feeling tendency that the application's embodiment provides;
Figure 22 is the device structure schematic diagram providing service based on user feeling tendency that another embodiment of the application provides;
Figure 23 is the device structure schematic diagram providing service based on user feeling tendency that another embodiment of the application provides;
Figure 24 is the device structure schematic diagram providing service based on user feeling tendency that another embodiment of the application provides;
Figure 25 is the device structure schematic diagram providing service based on user feeling tendency that another embodiment of the application provides.
Embodiment
Below in conjunction with drawings and Examples, the embodiment of the application is described in further detail.Following examples for illustration of the application, but are not used for limiting the scope of the application.
Those skilled in the art understand, in the embodiment of the application, the size of the sequence number of following each step does not also mean that the priority of execution sequence, and the execution sequence of each step should be determined with its function and internal logic, and should not form any restriction to the implementation process of the embodiment of the present application.
In addition, terms such as " first ", " second " in the application, only for distinguishing different step, equipment or module etc., neither representing any particular technology implication, also not representing the inevitable logical order between them.
Fig. 1 is the method setting up emotion model described in the application's embodiment, and see Fig. 1, described method comprises:
S100: determine at least one training text;
S120: carry out pre-service to described at least one training text, determines the emotion vector of each training text of word vector sum of each training text;
S140: according to the issuing time of the described word vector of each training text, the described emotion vector of each training text and each training text, determine the model that user feeling develops in time.
Optionally, in an embodiment of the application, determine at least one training text in above-mentioned S100, can comprise: obtain the content issued in a period of time from internet, as training text.Certainly, also can be obtain training text by other mode, such as, user manually imports at least one text as training text, and the embodiment of the application is not construed as limiting this.Foregoing can comprise: model of news, social networks speech (blog or microblogging etc.) or network forum etc.
After obtaining training text, just can carry out pre-service, determine the emotion vector (S120) of each training text of word vector sum of each training text.
Optionally, above-mentioned word vector can be effective word of each training text.Such as, the word vector of each training text can be expressed as:
d＝{w1,w2,w3,w1,w4,w5,w5,w2…}。
Optionally, above-mentioned emotion vector can be at least one Sentiment orientation produced after user has read each training text in advance.Such as, the emotion vector of each training text can be expressed as:
Wherein, n1, n2, n3, n4 ... the quantity producing this Sentiment orientation can be represented, exemplary, the quantity of above-mentioned a certain Sentiment orientation, can be the statistical magnitude producing this Sentiment orientation, or, can also be the normalization quantity that this Sentiment orientation produces, such as, this normalization quantity can be that the statistical magnitude that this Sentiment orientation produces accounts for the ratio of all Sentiment orientation statistical magnitudes.
For text vector, under normal circumstances, in a text, after getting rid of some insignificant words (such as, the word of " " " " and so on), remaining has been exactly effective word.
And for emotion vector, user is after reading training text, the option that can represent oneself Sentiment orientation can be selected by interactive interface as shown in Figure 2, therefore just can add up according to the data of user feedback, obtain the emotion vector of each training text.Or, after user's reading training text, the comment of oneself can be delivered by forms such as words, the comment of user is classified, just can obtain the emotion vector of each training text.
In the application's embodiment, after determining the emotion vector of each training text of word vector sum of each training text, by the vectorial and issuing time binding with the emotion of the training text of each training text of each the effective word in the word vector of each training text, obtain a tuple of training text: if training text d has Nd word, so the tuple of training text d also Nd is had.In all tuples of a training text, emotion vector all identical with issuing time t.Above-mentioned issuing time can specific to " hour " or " day ", can certainly granularity larger, such as, specific to " moon " or specific to " year " etc., the embodiment of the application does not do concrete restriction to this.
In another embodiment of the application, can divide according to issuing time to training text, belong to the training text in section at the same time, issuing time can be thought identical.Set up submodel separately for the training text belonged in section at the same time, then the user feeling described in the application just can comprise the submodel of multiple different time sections to the model of time-evolution.In each submodel, the issuing time due to training text is identical, and therefore, the tuple obtained can only include effective word and emotion vector, that is:
In the embodiment of the application, suppose that text is made up of all kinds of theme, " theme " herein represents a concept, an aspect, image, a bucket can be thought in theme, and the inside has filled the higher tuple of probability of occurrence, and these tuples and this theme have very strong correlativity.By " theme " this intermediary, just text and tuple can be connected.
Therefore, in the embodiment of the application, the probability that some texts produce a certain tuple can be obtained by " text is certain theme with certain probability selection, and this theme creates a certain tuple with certain probability " such process.That is:
P (tuple | text)=Σ themep (tuple | theme) × P (theme | text)
Each training text is added up, determines the number of times that in certain training text, some tuples occurs, just can obtain the P (tuple | training text) based on training text.Then by an algorithm, P (tuple | theme) and P (theme | training text) is obtained.According to training the P (tuple | theme) that obtains and P (theme | training text), just can determine that arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple.
In one alternate embodiment, as shown in Figure 3 a, according to the described word vector of each training text, the described emotion vector of each training text and the issuing time of each training text in above-mentioned S140, determine to comprise the model that user feeling develops in time:
S141: the emotion issuing time that is vectorial and i-th training text of each the effective word in the word vector of i-th training text and i-th training text is bound, obtains multiple tuples of i-th training text;
S142: add up i-th training text, determines the probability P (tuple r| training text i) occurring tuple r in described i-th training text;
S143: according to described P (tuple r| training text i), by an algorithm, determine the probability P (theme k| training text i) of described i-th training text choosing a topic k and the probability P (tuple r| theme k) of described theme k generation tuple r;
S144: for according to described P (theme k| training text i) and described P (tuple r| theme k), determines that arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple.
In another embodiment, as shown in Figure 3 b, according to the described word vector of each training text, the described emotion vector of each training text and the issuing time of each training text in above-mentioned S140, determine to comprise the model that user feeling develops in time:
S145: the emotion vector of each the effective word in the word vector of each training text and each training text is bound, obtains multiple tuples of each training text;
S146: add up the training text that at least one time period is issued, determines the probability P (tuple r| training text i) occurring tuple r in i-th training text issued in described at least one time period;
S147: according to described P (tuple r| training text i), by an algorithm, determine the probability P (theme k| training text i) of described i-th training text choosing a topic k and the probability P (tuple r| theme k) of described theme k generation tuple r;
S148: according to described P (theme k| training text i) and described P (tuple r| theme k), between determining at least for the moment, in section, arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple.
Optionally, above-mentioned algorithm, can be the algorithm based on sampling, or also can be variation EM (Expectation-maximization, expectation maximization) algorithm.Algorithm based on sampling passes through the sample collecting Posterior distrbutionp, tries to achieve the approximate of Posterior distrbutionp with the distribution of sample, and the common algorithm based on sampling is such as based on the algorithm of gibbs sampler (Gibbs Sampling).And variation EM algorithm first supposes the parameterized distribution of gang on concealed structure, then upgrade searching and the immediate distribution of Posterior distrbutionp by variation thought iteration.Below for the algorithm based on gibbs sampler, introduce two kinds of processes of Confirming model in the embodiment of the present application.
(1) suppose a training text collection D, have m training text d1, d2, d3 ... dm, issuing time is respectively: t1, t2, t3 ... tm, n theme Z1, Z2, Z3 ... Zn.
Each training text concentrated by training text changes into the form of tuple:
Time initial: with the probability of equalization or random each tuple for each training text gives a theme, exemplary, as shown in table 1:
Table 1
Giving situation according to initial theme, is the tuple in i-th training text di calculate theme probability, that is: get rid of the theme assignment of this tuple, according to the theme assignment of other all tuples (comprising other tuples of i-th training text and all tuples of other training texts), estimate tuple in text di be endowed the probability of theme Zk
Obtain this tuple after belonging to the probability of each theme, be this tuple again according to these probability give a theme.
Then use the same method and upgrade the theme of next tuple, until find distribution P (tuple | the theme) convergence of the tuple under theme distribution P under each training text (theme | training text) and each theme, algorithm stops.Exemplary, after algorithm stops, the situation that theme is given is as shown in table 2:
Table 2
Situation about then just can give according to up-to-date theme, carry out several quantitative statistics, determines that arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple, such as, can add up:
There is the probability P (theme | text) of arbitrary theme in arbitrary text;
Arbitrary theme uses the probability P (effective word | theme) of arbitrary effective word;
Arbitrary theme produces the probability P (Sentiment orientation | theme) of arbitrary Sentiment orientation; And
Arbitrary theme results from the probability P (time | theme) of any one time.
Above-mentioned probability distribution, just can think the output of the model that the user feeling described in the embodiment of the present application develops in time.
(2) multiple training text subset D is supposed t1, D t2, D t3d tm, each training text subset comprises at least one training text, and each training text subset issuing time is respectively: t1, t2, t3 ... tm, has n1 theme in training text subset D 1, there is n2 theme in training text subset D 2, in training text subset D 3, have n3 theme ...Wherein, theme corresponding to each training text subset can be different, and the quantity of theme also can be different.
Owing to being identical according to the process of each training text subset Modling model, therefore, just only describe in detail for a training text subset below.
Each training text in certain training text subset is changed into the form of tuple:
Similar with a upper process, each tuple at random for each training text time initial gives a theme; Giving situation according to initial theme, is a certain tuple in some training text di calculate theme probability; Obtain this tuple after belonging to the probability of each theme, be this tuple again according to this probability distribution give a theme.Then use the same method and upgrade the theme of next tuple, until find distribution P (tuple | the theme) convergence of the tuple under theme distribution P under each training text (theme | text) and each theme, algorithm stops.
Situation about then just can give according to up-to-date theme, carry out several quantitative statistics, and between determining at least for the moment, in section, arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple, such as, can add up:
There is the probability P (theme | text) of arbitrary theme in arbitrary text;
Arbitrary theme uses the probability P (effective word | theme) of arbitrary effective word; And
Arbitrary theme produces the probability P (Sentiment orientation | theme) of arbitrary Sentiment orientation.
Because training text subset divided according to the time, therefore, after setting up submodel for each training text subset, the model of user feeling Temporal Evolution can just be obtained.
Therefore, in the application's embodiment, determine in above-mentioned S144 or S148 that arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple, can comprise:
A: determine that the probability of arbitrary theme appears in arbitrary text;
B: determine that arbitrary theme uses the probability of arbitrary effective word;
C: determine that arbitrary theme produces the probability of arbitrary Sentiment orientation; And
D: determine that arbitrary theme results from the probability of arbitrary time.
Adopt the method setting up emotion model described in the embodiment of the present application, can predict the differentiation in time of the emotion of user, compensate for the defect of prior art.
Further, according to the emotion model that the embodiment of the present application is set up, more practical application can also be had.
Scene one
In this scene, have one or more text to be assessed, if need this text to be assessed of prediction to issue, which type of Sentiment orientation military order reader produces.
As shown in Figure 4, an embodiment of the application provides a kind of method that user feeling is inclined to of predicting, see Fig. 4, described method comprises:
S400: determine at least one text to be assessed;
S420: the model developed in time according to a user feeling, determines that the arbitrary text in described at least one text to be assessed makes user produce the probability of at least one Sentiment orientation.
Adopt the method for the prediction user feeling tendency described in the application, can predict not delivering user's feelings tendency that text may cause, compensate for the defect of prior art.
Optionally, the model that the described user feeling in S420 develops in time can be the model that user feeling that the method establishment described by the application's aforementioned process (1) is got up develops in time.Training text for model foundation can be collected in advance and obtain.
After determining one section or many sections of texts to be assessed, just can carry out pre-service to described one or more text to be assessed, described one or more text to be assessed is resolved into the form of aforesaid tuple, input the model that described user feeling develops in time, and then according to the result that model exports, any one text d ' obtained in text to be assessed makes user produce the probability of certain Sentiment orientation e, that is: P (e|d ').
Exemplary,
P ( e | d ′ ) = Σ k = 1 N topic [ P ( e | Z k ) × P ( Z k | d ′ ) ] ---formula 1
Wherein, N topicrepresent total quantity of theme, P (Z k| d ') represent that theme Z appears in text d ' to be assessed kprobability, P (e|Z k) represent theme Z kmake user produce a feeling and be inclined to the probability of e.
The model that the user feeling described in the method setting up emotion model according to the application's previous embodiment develops in time, above-mentioned can be provided by the Output rusults of model.Therefore, the known variables in formula 1 is P (Z k| d ').
According to Bayesian formula:
P ( Z k | d ′ ) = P ( d ′ | Z k ) × P ( Z k ) P ( d ′ ) ∝ P ( d ′ | Z k ) × P ( Z k ) ---formula 2
Wherein, symbol ∝ represents " being proportional to ", P (Z k) be the theme Z kprior probability, this prior probability can be the empirical value of forefathers, also can be to obtain according to the training text in model process of establishing: wherein, N drepresent the quantity of training text.Therefore, the known variables in formula 2 be P (d ' | Z k).
Suppose the model set up according to said process (1), then because this assessment text d ' can change at least one tuple form, wherein, w ' ∈ d ', therefore:
---formula 3
Wherein, symbol ∈ represents " belonging to ", represent the emotion vector of the Sentiment orientation composition that text d ' to be assessed may make user produce, t ' represents the release time being anticipated of text to be assessed, and w ' represents effective word of text to be assessed.
So far, all variablees in formula 3 can be provided by the Output rusults of model.The Output rusults of combination model, and above-mentioned formula 1,2,3, just can obtain P (e|d '), also just obtain text d ' to be assessed and make user produce the probability of certain Sentiment orientation e.
Scene two
In this scene, the text issued in the past period can be collected, therefrom analyze the abnormal part of user feeling tendency, and follow the trail of the time of this abnormal emotion tendency generation.
As shown in Figure 5, in another embodiment of the application, provide a kind of method monitoring public sentiment, see Fig. 5, described method can comprise:
S500: determine text to be assessed, wherein, described text to be assessed is at least included in the first text set issued in first time period and the second text set issued within the second time period;
S520: the model developed in time according to a user feeling, at least determine that described first text set makes user produce the probability distribution of each Sentiment orientation in described first time period, and described second text set makes user produce the probability distribution of each Sentiment orientation in described second time period;
S540: at least make user produce the probability distribution of each Sentiment orientation according to described first text set in described first time period, and described second text set makes user produce the probability distribution of each Sentiment orientation in described second time period, determine that at least one abnormal emotion is inclined to;
S560: determine that each abnormal emotion tendency in described at least one abnormal emotion tendency results from the probability of arbitrary time.
Adopt the method for the monitoring public sentiment described in the application, excavation and the analysis of the degree of depth can be carried out magnanimity public sentiment, determine the Sentiment orientation differentiation in time of user, compensate for the defect of prior art.
Optionally, the model that in described S520, user feeling develops in time can be the model that user feeling that the method establishment described by said process (2) is got up develops in time.
In this scene, first the time period of text to be assessed will be divided, then, according to the time period, pre-service is carried out to text to be assessed, the emotion vector of each the effective word in the word vector of each text to be assessed and text to be assessed is bound, obtains the tuple of text to be assessed, input in the submodel corresponding with the issuing time of text to be assessed self.
According to the submodel of different time sections, just can determine that the text to be assessed that different time sections is issued makes user produce the probability of certain Sentiment orientation e within this time period:
P ( e | D t 1 ) ∝ Π d ′ ∈ D t 1 P ( e | d ′ )
P ( e | D t 2 ) ∝ Π d ′ ∈ D t 2 P ( e | d ′ )
P ( e | D t 3 ) ∝ Π d ′ ∈ D t 3 P ( e | d ′ )
……
Then, just can according to P (e|D t1), P (e|D t2), P (e|D t3) ... determine the fluctuation situation of this Sentiment orientation e.Such as, can calculate the text issued in the t1 time period makes user produce the probability distribution of each Sentiment orientation: P (e 1| D t1), P (e 2| D t1), P (e 3| D t1), P (e 4| D t1), P (e 5| D t1) ..., and the text issued in the t2 time period makes user produce the probability distribution of each Sentiment orientation: P (e 1| D t2), P (e 2| D t2), P (e 3| D t2), P (e 4| D t2), P (e 5| D t2) ..., then calculate the distance of these two probability distribution, such as, COS distance or non symmetrical distance, thus this connects in time period to determine time period t 1 and time period t 2 according to the distance of above-mentioned probability distribution, the fluctuation situation of each Sentiment orientation.
If the fluctuation situation of certain Sentiment orientation has exceeded threshold value, just can think that this Sentiment orientation is abnormal emotion tendency.
Determine that certain abnormal emotion inclines e ito the probability resulting from certain time, namely determine P (t|e i):
P ( t | e ) = Σ k = 1 N topic [ P ( t | Z k ) × P ( Z k | e ) ] ∝ Σ k = 1 N topic [ P ( t | Z k ) × P ( e | Z k ) × P ( Z k ) ]
Wherein, N topicrepresent total quantity of theme.
Scene three
In this scene, the text that user read can be collected, therefrom analyze the text that user read, the probability distribution on emotional semantic classification, and based on this for user provides service.
As shown in Figure 6, in another embodiment of the application, provide a kind of method providing service based on user feeling tendency, see Fig. 6, described method can comprise:
S600: determine text to be assessed, wherein, described text to be assessed is at least one text that a user read;
S620: the model developed in time according to a user feeling, determines that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation;
S640: make described user produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service.
Adopt the method that service is provided based on user feeling tendency described in the application, depth analysis can be carried out to user's feelings tendency, and provide accordingly and serve targetedly, compensate for the defect of prior art.
Optionally, the model that in described S620, user feeling develops in time, can be the model that user feeling that the method establishment described by said process (1) is got up develops in time, or also can be the model that user feeling that the method establishment described by said process (2) is got up develops in time.
In this scene, first each text to be assessed is processed, obtain the tuple of each text to be assessed, input described model, finally obtain P (e|d ').Obtain the process of P (e|d '), all have a detailed description in above-mentioned scene one or scene two, repeat no more herein.
User is made to produce the probability of each Sentiment orientation according to each text to be assessed, just can obtain after all texts to be assessed of user make user produce the probability distribution of each Sentiment orientation, optionally, based on described probability distribution, the service that can provide for described user can comprise at least one in following service:
A. at least one content is pushed for described user, or request network side is that described user pushes at least one content, wherein, the probability distribution that the text described to be assessed that described at least one content makes described user produce obtaining in the probability distribution of each Sentiment orientation and above-mentioned S620 makes described user produce each Sentiment orientation matches.Exemplary, at least one content of the above-mentioned user of being pushed to can comprise: at least one news; Or post at least one network forum; Or at least one social networks speech.Certainly, can also be other forms of content, embodiments of the invention limit this.
B. be described user's commending friends, or request network side is described user's commending friends, wherein, the probability distribution that the text described to be assessed that the text that described good friend read makes described good friend produce obtaining in the probability distribution of each Sentiment orientation and above-mentioned S620 makes described user produce each Sentiment orientation matches.
Present invention also provides a kind of device setting up emotion model, see Fig. 7, the described device setting up emotion model can comprise:
Determination module 700, for determining at least one training text;
Pretreatment module 720, for carrying out pre-service to described at least one training text, determines the emotion vector of each training text of word vector sum of each training text;
MBM 740, for the described word vector according to each training text, the described emotion vector of each training text and the issuing time of each training text, determines the model that user feeling develops in time.
Optionally, as shown in Figure 8, described pretreatment module 720 can comprise:
Word vector pre-service submodule 721, for filtering the meaningless word of each training text, determines effective word of each training text;
Emotion vector pre-service submodule 722, for obtaining at least one Sentiment orientation produced after user reads each training text, and adds up the quantity that in described at least one Sentiment orientation, often kind of Sentiment orientation produces.
Optionally, as shown in Figure 9, described MBM 740 can comprise:
First tuple determination submodule 741, for the issuing time binding by the emotion of each the effective word in the word vector of i-th training text and i-th training text vector and i-th training text, obtains the tuple of i-th training text;
First statistics submodule 742, for adding up i-th training text, determines the probability P (tuple r| training text i) occurring tuple r in described i-th training text;
First training submodule 743, for according to described P (tuple r| training text i), by an algorithm, determine the probability P (theme k| training text i) of i-th training text choosing a topic k and the probability P (tuple r| theme k) of described theme k generation tuple r;
First output sub-module 744, for according to described P (theme k| training text i) and described P (tuple r| theme k), determines that arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple.
Optionally, as shown in Figure 10, described MBM 740 can comprise:
Second tuple determination submodule 745, for the emotion vector binding by each the effective word in the word vector of each training text and each white silk text of instruction, obtains the tuple of each training text;
Second statistics submodule 746, adds up for the training text issued at least one time period, determines the probability P (tuple r| training text i) occurring tuple r in i-th training text issued in described at least one time period;
Second training submodule 747, for according to described P (tuple r| training text i), by an algorithm, determine the probability P (theme k| training text i) of described i-th training text choosing a topic k and the probability P (tuple r| theme k) of described theme k generation tuple r;
Second output sub-module 748, for according to described P (theme k| training text i) and described P (tuple r| theme k), between determining at least for the moment, in section, arbitrary text produces the approximate Posterior distrbutionp P (tuple | text) of arbitrary tuple.
Optionally, as shown in figure 11, described first output sub-module 744 or described second output sub-module 748, can comprise:
First output unit 749-1, for determining that the probability of arbitrary theme appears in arbitrary text;
Second output unit 749-2, for determining that arbitrary theme uses the probability of arbitrary effective word;
3rd output unit 749-3, for determining that arbitrary theme produces the probability of arbitrary Sentiment orientation; And
4th output unit 749-4, for determining that arbitrary theme results from the probability of arbitrary time.
Adopt the device setting up emotion model described in the embodiment of the present application, can predict the differentiation in time of the emotion of user, compensate for the defect of prior art.
Those skilled in the art can be well understood to, for convenience and simplicity of description, the specific works process setting up the device of emotion model of foregoing description, the aforementioned corresponding process setting up the embodiment of the method for emotion model of REFERENCE TO RELATED can describe, does not repeat them here.
Present invention also provides a kind of equipment that user feeling is inclined to of predicting, as shown in figure 12, the equipment of described prediction user feeling tendency can comprise:
One assessment text determining device 1200, for determining at least one text to be assessed;
One prediction unit 1220, for the model developed in time according to a user feeling, determines that the arbitrary text in described at least one text to be assessed makes user produce the probability of at least one Sentiment orientation.
Optionally, as shown in figure 13, described prediction unit 1220 can comprise:
Processing module 1221, for carrying out pre-service to described at least one text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
Tuple determination module 1222, for the issuing time binding by the emotion of each the effective word in the word vector of each text to be assessed and each text to be assessed vector and each text to be assessed, obtains the tuple of each text to be assessed;
Load module 1223, for the tuple of each text to be assessed described being inputted the model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that the arbitrary text in described at least one text to be assessed makes user produce the probability of at least one Sentiment orientation.
Optionally, as shown in figure 14, the equipment of described prediction user feeling tendency can also comprise:
One device 1240 setting up emotion model, for setting up the model that described user feeling develops in time.
Optionally, as shown in figure 15, the described device 1240 setting up emotion model can comprise:
Determination module 1241, for determining at least one training text;
Pretreatment module 1242, for carrying out pre-service to described at least one training text, determines the emotion vector of each training text of word vector sum of each training text;
MBM 1243, for the described word vector according to each training text, the described emotion vector of each training text and the issuing time of each training text, determines the model that user feeling develops in time.
Optionally, described pretreatment module 1242 can be the pretreatment module 720 as shown in Figure 8 in embodiment, repeats no more herein.
Optionally, described MBM 1243 can be the MBM 740 as shown in Figure 9 in embodiment, repeats no more herein.
Adopt the equipment of the prediction user feeling tendency described in the application, can predict not delivering the user feeling tendency that text may cause, compensate for the defect of prior art.
Present invention also provides a kind of equipment monitoring public sentiment, as shown in figure 16, the equipment of described monitoring public sentiment can comprise:
One assessment text determining device 1600, for determining text to be assessed, wherein, described text to be assessed is at least included in the first text set issued in first time period and the second text set issued within the second time period;
One emotion probability distribution determining device 1620, for the model developed in time according to a user feeling, at least determine that described first text set makes user produce the probability distribution of each Sentiment orientation in described first time period, and described second text set makes user produce the probability distribution of each Sentiment orientation in described second time period;
One abnormal emotion determining device 1640, for at least making user produce the probability distribution of each Sentiment orientation according to described first text set in described first time period, and described second text set makes user produce the probability distribution of each Sentiment orientation in described second time period, determine that at least one abnormal emotion is inclined to;
One time probability distribution determining device 1660, for determining that each abnormal emotion tendency in described at least one abnormal emotion tendency results from the probability of arbitrary time.
Optionally, as shown in figure 17, described emotion probability distribution determining device 1620, can comprise:
Processing module 1621, for carrying out pre-service to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
Tuple determination module 1622, for by the binding of the emotion of each the effective word in the word vector of each text to be assessed and each text to be assessed vector, obtains the tuple of each text to be assessed;
Load module 1623, for the tuple of each text to be assessed being inputted submodel corresponding with the issuing time of each text to be assessed described in model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, at least determine that described first text concentrates on described first time period and makes user produce the probability distribution of each Sentiment orientation, and described second text set makes user produce the probability distribution of each Sentiment orientation in described second time period.
Optionally, as shown in figure 18, described abnormal emotion determining device 1640, can comprise:
Sentiment orientation fluctuation determination module 1641, at least determining that the probability distribution that the text issued in first time period makes user produce each Sentiment orientation and the text issued in the second time period make user produce the distance of the probability distribution of each Sentiment orientation;
Abnormal emotion determination module 1642, the probability distribution producing each Sentiment orientation for making user according to the text issued in described first time period and the text issued in the second time period make user produce the distance of the probability distribution of each Sentiment orientation, determine that at least one abnormal emotion is inclined to.
Optionally, as shown in figure 19, the equipment of described monitoring public sentiment can also comprise:
One device 1680 setting up emotion model, for setting up the model that described user feeling develops in time.
Optionally, as shown in figure 20, the described device 1680 setting up emotion model can comprise:
Determination module 1681, for determining at least one training text;
Pretreatment module 1682, for carrying out pre-service to described at least one training text, determines the emotion vector of each training text of word vector sum of each training text;
MBM 1683, for the described word vector according to each training text, the described emotion vector of each training text and the issuing time of each training text, determines the model that user feeling develops in time.
Optionally, described pretreatment module 1682 can be the pretreatment module 720 as shown in Figure 8 in embodiment, repeats no more herein.
Optionally, described MBM 1683 can be the MBM 740 as shown in Figure 10 in embodiment, repeats no more herein.
Adopt the equipment of the monitoring public sentiment described in the application, excavation and the analysis of the degree of depth can be carried out magnanimity public sentiment, determine the Sentiment orientation differentiation in time of user, compensate for the defect of prior art.
One of the application embodiment still provides a kind of equipment providing service based on user feeling tendency, and see Figure 21, described tendency based on user feeling provides the equipment of service to comprise:
One assessment text determining device 2100, for determining text to be assessed, wherein, described text to be assessed is at least one text that a user read;
One emotion probability distribution determining device 2120, for the model developed in time according to a user feeling, determines that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation;
One service providing apparatus 2140, for making described user produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service.
Optionally, as shown in figure 22, described emotion probability distribution determining device 2120, can comprise:
First processing module 2121, for carrying out pre-service to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
First tuple determination module 2122, for the issuing time binding by the emotion of each the effective word in the word vector of each text to be assessed and each text to be assessed vector and each text to be assessed, obtain the tuple of each text to be assessed;
First load module 2123, for the tuple of each text to be assessed being inputted the model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that described text to be assessed makes user produce the probability distribution of each Sentiment orientation.
Optionally, as shown in figure 23, described emotion probability distribution determining device 2120, can comprise:
Second processing module 2124, for carrying out pre-service to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
Second tuple determination module 2125, for by the binding of the emotion of each the effective word in the word vector of each text to be assessed and each text to be assessed vector, obtains the tuple of each text to be assessed;
Second load module 2126, for the tuple of each text to be assessed being inputted submodel corresponding with the issuing time of each text to be assessed in model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that described text to be assessed makes user produce the probability distribution of each Sentiment orientation.
Optionally, as shown in figure 24, described service providing apparatus 2140, can comprise:
First service provides module 2141, for pushing at least one content for described user, or request network side is that described user pushes at least one content, wherein, described at least one content make described user produce probability distribution that the probability distribution of each Sentiment orientation and described text to be assessed make described user produce each Sentiment orientation matches; And/or
Second service provides module 2142, for being described user's commending friends, or request network side is described user's commending friends, wherein, the text that described good friend read makes described good friend produce, and probability distribution that the probability distribution of each Sentiment orientation and described text to be assessed make described user produce each Sentiment orientation matches.
Optionally, as shown in figure 25, described tendency based on user feeling provides the equipment of service to comprise:
One device 2160 setting up emotion model, for setting up the model that described user feeling develops in time.
Optionally, the described device 2160 setting up emotion model can be the device setting up emotion model as described in accompanying drawing illustrated embodiment arbitrary in Fig. 7 to Figure 11, repeats no more herein.
Adopt the equipment that service is provided based on user feeling tendency described in the application, depth analysis can be carried out to user's feelings tendency, and provide accordingly and serve targetedly, compensate for the defect of prior art.
Those of ordinary skill in the art can recognize, in conjunction with unit and the method step of each example of embodiment disclosed herein description, can realize with the combination of electronic hardware or computer software and electronic hardware.These functions perform with hardware or software mode actually, depend on application-specific and the design constraint of technical scheme.Professional and technical personnel can use distinct methods to realize described function to each specifically should being used for, but this realization should not think the scope exceeding the application.
If described function using the form of SFU software functional unit realize and as independently production marketing or use time, can be stored in a computer read/write memory medium.Based on such understanding, the part of the part that the technical scheme of the application contributes to prior art in essence in other words or this technical scheme can embody with the form of software product, this computer software product is stored in a storage medium, comprising some instructions in order to make a computer equipment (can be personal computer, controller, or the network equipment etc.) perform all or part of step of method described in each embodiment of the application.And aforesaid storage medium comprises: USB flash disk, portable hard drive, ROM (read-only memory) (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. various can be program code stored medium.
Above embodiment is only for illustration of the application; and the restriction not to the application; the those of ordinary skill of relevant technical field; when not departing from the spirit and scope of the application; can also make a variety of changes and modification; therefore all equivalent technical schemes also belong to the category of the application, and the scope of patent protection of the application should be defined by the claims.

Claims (10)

1. a method for service is provided based on user feeling tendency, it is characterized in that, comprising:
Determine text to be assessed, wherein, described text to be assessed is at least one text that a user read;
According to the model that a user feeling develops in time, determine that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation;
Described user is made to produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service.
2. the method for claim 1, is characterized in that, according to the model that a user feeling develops in time, determines that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation, comprising:
Pre-service is carried out to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
The emotion issuing time that is vectorial and each text to be assessed of each the effective word in the word vector of each text to be assessed and each text to be assessed is bound, obtains the tuple of each text to be assessed;
The tuple of each text to be assessed is inputted the model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that described text to be assessed makes user produce the probability distribution of each Sentiment orientation;
Or
Pre-service is carried out to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
The emotion vector of each the effective word in the word vector of each text to be assessed and each text to be assessed is bound, obtains the tuple of each text to be assessed;
The tuple of each text to be assessed is inputted submodel corresponding with the issuing time of each text to be assessed in the model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that described text to be assessed makes user produce the probability distribution of each Sentiment orientation.
3. method as claimed in claim 1 or 2, is characterized in that, describedly makes described user produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service, comprising:
For described user pushes at least one content, or request network side is that described user pushes at least one content, wherein, described at least one content make described user produce probability distribution that the probability distribution of each Sentiment orientation and described text to be assessed make described user produce each Sentiment orientation matches; And/or
For described user's commending friends, or request network side is described user's commending friends, wherein, the text that described good friend read makes described good friend produce, and probability distribution that the probability distribution of each Sentiment orientation and described text to be assessed make described user produce each Sentiment orientation matches.
4., as the method as described in arbitrary in claims 1 to 3, it is characterized in that, described method also comprises:
Set up the model that described user feeling develops in time.
5. method as claimed in claim 4, it is characterized in that, the described model set up described user feeling and develop in time, comprising:
Determine at least one training text;
Pre-service is carried out to described at least one training text, determines the emotion vector of each training text of word vector sum of each training text;
According to the issuing time of the described word vector of each training text, the described emotion vector of each training text and each training text, determine the model that user feeling develops in time.
6. provide an equipment for service based on user feeling tendency, it is characterized in that, described tendency based on user feeling provides the equipment of service to comprise:
One assessment text determining device, for determining text to be assessed, wherein, described text to be assessed is at least one text that a user read;
One emotion probability distribution determining device, for the model developed in time according to a user feeling, determines that described text to be assessed makes described user produce the probability distribution of each Sentiment orientation;
One service providing apparatus, for making described user produce the probability distribution of each Sentiment orientation according to described text to be assessed, for described user provides service.
7. equipment as claimed in claim 6, it is characterized in that, described emotion probability distribution determining device comprises:
First processing module, for carrying out pre-service to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
First tuple determination module, for the issuing time binding by the emotion of each the effective word in the word vector of each text to be assessed and each text to be assessed vector and each text to be assessed, obtains the tuple of each text to be assessed;
First load module, for the tuple of each text to be assessed being inputted the model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that described text to be assessed makes user produce the probability distribution of each Sentiment orientation;
Or
Second processing module, for carrying out pre-service to described text to be assessed, determines the emotion vector of each text to be assessed of word vector sum of each text to be assessed;
Second tuple determination module, for by the binding of the emotion of each the effective word in the word vector of each text to be assessed and each text to be assessed vector, obtains the tuple of each text to be assessed;
Second load module, for the tuple of each text to be assessed being inputted submodel corresponding with the issuing time of each text to be assessed in model that described user feeling develops in time, according to the Output rusults of the model that described user feeling develops in time, determine that described text to be assessed makes user produce the probability distribution of each Sentiment orientation.
8. equipment as claimed in claims 6 or 7, it is characterized in that, described service providing apparatus comprises:
First service provides module, for pushing at least one content for described user, or request network side is that described user pushes at least one content, wherein, described at least one content make described user produce probability distribution that the probability distribution of each Sentiment orientation and described text to be assessed make described user produce each Sentiment orientation matches; And/or
Second service provides module, for being described user's commending friends, or request network side is described user's commending friends, wherein, the text that described good friend read makes described good friend produce, and probability distribution that the probability distribution of each Sentiment orientation and described text to be assessed make described user produce each Sentiment orientation matches.
9., as the equipment as described in arbitrary in claim 6 to 8, it is characterized in that, the equipment of described monitoring public sentiment also comprises:
One device setting up emotion model, for setting up the model that described user feeling develops in time.
10. equipment as claimed in claim 9, it is characterized in that, the described device setting up emotion model comprises:
Determination module, for determining at least one training text;
Pretreatment module, for carrying out pre-service to described at least one training text, determines the emotion vector of each training text of word vector sum of each training text;
MBM, for the described word vector according to each training text, the described emotion vector of each training text and the issuing time of each training text, determines the model that user feeling develops in time.
CN201410773678.3A 2014-12-12 2014-12-12 The method and apparatus for being inclined to the service of offer based on user feeling Active CN104504032B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410773678.3A CN104504032B (en) 2014-12-12 2014-12-12 The method and apparatus for being inclined to the service of offer based on user feeling

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410773678.3A CN104504032B (en) 2014-12-12 2014-12-12 The method and apparatus for being inclined to the service of offer based on user feeling

Publications (2)

Publication Number Publication Date
CN104504032A true CN104504032A (en) 2015-04-08
CN104504032B CN104504032B (en) 2019-03-01

Family

ID=52945430

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410773678.3A Active CN104504032B (en) 2014-12-12 2014-12-12 The method and apparatus for being inclined to the service of offer based on user feeling

Country Status (1)

Country Link
CN (1) CN104504032B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108460111A (en) * 2018-02-08 2018-08-28 合肥工业大学 Personal character modeling based on dialogue and generation method and device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103365867A (en) * 2012-03-29 2013-10-23 腾讯科技（深圳）有限公司 Method and device for emotion analysis of user evaluation
CN103559233A (en) * 2012-10-29 2014-02-05 中国人民解放军国防科学技术大学 Extraction method for network new words in microblogs and microblog emotion analysis method and system
TW201416887A (en) * 2012-10-30 2014-05-01 Univ Ming Chuan Methods for sentimental analysis of news text
CN104063399A (en) * 2013-03-22 2014-09-24 杭州金弩信息技术有限公司 Method and system for automatically identifying emotional probability borne by texts

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103365867A (en) * 2012-03-29 2013-10-23 腾讯科技（深圳）有限公司 Method and device for emotion analysis of user evaluation
CN103559233A (en) * 2012-10-29 2014-02-05 中国人民解放军国防科学技术大学 Extraction method for network new words in microblogs and microblog emotion analysis method and system
TW201416887A (en) * 2012-10-30 2014-05-01 Univ Ming Chuan Methods for sentimental analysis of news text
CN104063399A (en) * 2013-03-22 2014-09-24 杭州金弩信息技术有限公司 Method and system for automatically identifying emotional probability borne by texts

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108460111A (en) * 2018-02-08 2018-08-28 合肥工业大学 Personal character modeling based on dialogue and generation method and device
CN108460111B (en) * 2018-02-08 2020-10-16 合肥工业大学 Personal character modeling and generating method and device based on conversation

Also Published As

Publication number Publication date
CN104504032B (en) 2019-03-01

Similar Documents

Publication Publication Date Title
Rudra et al. Identifying sub-events and summarizing disaster-related information from microblogs
Lakkaraju et al. What's in a name? understanding the interplay between titles, content, and communities in social media
CN109145216A (en) Network public-opinion monitoring method, device and storage medium
Shi et al. Sentiment analysis of Chinese microblogging based on sentiment ontology: a case study of ‘7.23 Wenzhou Train Collision’
CN112749326B (en) Information processing method, information processing device, computer equipment and storage medium
JP2019519019A (en) Method, apparatus and device for identifying text type
US8489626B2 (en) Method and apparatus for recommending a short message recipient
CN103399891A (en) Method, device and system for automatic recommendation of network content
CN103870001A (en) Input method candidate item generating method and electronic device
CN104731958A (en) User-demand-oriented cloud manufacturing service recommendation method
CN109857869B (en) Ap incremental clustering and network element-based hot topic prediction method
CN111522886B (en) Information recommendation method, terminal and storage medium
CN105095311A (en) Method, device and system for processing promotion information
Tindall et al. Big data, computational social science, and other recent innovations in social network analysis
CN104516873A (en) Method and device for building emotion model
CN102156747A (en) Method and device for forecasting collaborative filtering mark by introduction of social tag
CN113254652A (en) Social media posting authenticity detection method based on hypergraph attention network
CN112215629B (en) Multi-target advertisement generating system and method based on construction countermeasure sample
Susilawati Public services satisfaction based on sentiment analysis: Case study: Electrical services in Indonesia
CN114138968A (en) Network hotspot mining method, device, equipment and storage medium
Campbell et al. Content+ context networks for user classification in twitter
Shrivatava et al. Opinion mining of real time twitter tweets
CN104503959A (en) Method and equipment for predicting user emotion tendency
Kanavos et al. Fuzzy information diffusion in twitter by considering user’s influence
CN104504032A (en) Method and equipment for providing service upon user emotion tendencies

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant