CN108595564A - Media friendliness appraisal procedure, device and computer readable storage medium - Google Patents
Media friendliness appraisal procedure, device and computer readable storage medium Download PDFInfo
- Publication number
- CN108595564A CN108595564A CN201810330401.1A CN201810330401A CN108595564A CN 108595564 A CN108595564 A CN 108595564A CN 201810330401 A CN201810330401 A CN 201810330401A CN 108595564 A CN108595564 A CN 108595564A
- Authority
- CN
- China
- Prior art keywords
- media
- friendliness
- text
- index
- related text
- 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
- 238000000034 method Methods 0.000 title claims abstract description 89
- 238000003860 storage Methods 0.000 title claims abstract description 13
- 230000008451 emotion Effects 0.000 claims abstract description 72
- 238000004458 analytical method Methods 0.000 claims abstract description 47
- 238000006243 chemical reaction Methods 0.000 claims abstract description 14
- 230000007935 neutral effect Effects 0.000 claims description 14
- 230000002996 emotional effect Effects 0.000 claims description 9
- 230000002045 lasting effect Effects 0.000 claims description 9
- 238000010801 machine learning Methods 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 6
- 230000007774 longterm Effects 0.000 abstract description 17
- 238000004364 calculation method Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 238000004590 computer program Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 239000000969 carrier Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 230000033764 rhythmic process Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of media friendliness appraisal procedure, device and computer readable storage mediums, are related to the analysis of public opinion technical field, method includes：Obtain target medium publication with the relevant multiple related texts of public sentiment object；Feeling polarities analysis is carried out to each related text in multiple related texts, obtains the feeling polarities value of each related text；The feeling polarities value of each related text is converted into the corresponding media friendliness index of each related text respectively, and constitutes media friendliness exponential time sequence；Media friendliness exponential time sequence is smoothed, the media friendliness exponential time sequence after obtaining smoothly；Media friendliness based on the media friendliness exponential time sequence estimation target medium after smooth.The embodiment of the present invention can realize the conversion of the long-term attitude from text emotion to media, and then realize that accurately assessment target medium is to the attitude steady in a long-term of public sentiment object.
Description
Technical field
The present invention relates to the analysis of public opinion technical field, more particularly to a kind of media friendliness appraisal procedure, device and calculating
Machine readable storage medium storing program for executing.
Background technology
To certain hot spots, focal issue in actual life, traditional media, leader of opinion and general public can be delivered
With tendentious speech and viewpoint.Public sentiment, speech and sight are known as to some event or phenomenon published utterances and viewpoint
Point hold and publisher is known as public sentiment main body, and it is public sentiment object to be paid close attention to the object commented on by public sentiment main body, speech with see
The media (including the network media and traditionally on paper media) of point publication are then public sentiment carriers.Public sentiment monitoring is pair and public sentiment object phase
The behavior that the speech and viewpoint that the public sentiment main body of pass is delivered are monitored, analyzed and predicted.
Media have the characteristics that profession gather and edit, public publication, audient it is extensive, be the main source that people obtain information, together
When also to audient have important influence, therefore be public sentiment monitoring main object.Continuous with internet is popularized, and network is
Through the basic tool as work and life, media onlineization also becomes big trend, and the influence power of emerging network media is more next
Bigger, traditionally on paper media also issue the network edition simultaneously mostly.Media network accelerates the rhythm of media distribution and propagation, net
A large amount of appearance of network public sentiment are that public sentiment monitors and analysis proposes new challenge and opportunity.From the angle of technology, network public-opinion prison
Survey refers to integrating internet information acquisition technology and information intelligent treatment technology, by being captured automatically to internet mass information,
Automatic taxonomic clustering, topic detection, focus on special topic realize that the information such as network public-opinion monitoring and the Special Topics in Journalism tracking of user need
It asks, forms the analysis results such as bulletin, report, chart.
Public sentiment content not only delivers Outside View of the public sentiment main body to public sentiment object, further comprises the subjectivity of public sentiment main body
Emotion and mood.Text emotion analysis is also known as opinion mining, is to be analyzed the subjective texts with emotional color, located
Reason, the process of conclusion and reasoning.Initial sentiment analysis mainly for the word with emotional color analysis, e.g., " fine "
It is the word with commendation color, and " ugliness " is the word of color of conveying a derogatory sense.Later, related researcher was gradually from simple
The analysis and research of emotion word be transitioned into increasingly complex emotion sentence research and the research of emotion chapter；In general, main
The polarity for seeing this paper is divided into commendation and two class of derogatory sense or commendation, neutrality and derogatory sense three classes.
Text emotion analytical technology towards be single document, only cannot react some matchmaker from the Sentiment orientation of single document
Body treats the long-term attitude of public sentiment object.Media friendliness assessment target be observation media treat public sentiment object (such as some
Company) attitude steady in a long-term.Understand media friendliness and can be used to identify different tendentious media, to give in various degree
Concern.
Media friendliness is assessed and tracked however, it is found by the inventors that not yet disclosing result of study in the prior art.
Invention content
In order to solve problems in the prior art, an embodiment of the present invention provides a kind of media friendliness appraisal procedure and dresses
It sets, to assess media friendliness of the media to some public sentiment object, to understand its attitude steady in a long-term.
Specific technical solution provided in an embodiment of the present invention is as follows：
In a first aspect, a kind of media friendliness appraisal procedure is provided, the method includes the steps：
Obtain target medium publication with the relevant multiple related texts of public sentiment object；
Feeling polarities analysis is carried out to each related text in the multiple related text, obtains each related text
Feeling polarities value；
The feeling polarities value of each related text is converted into the corresponding media friendliness of each related text respectively
Index, and constitute media friendliness exponential time sequence；
The media friendliness exponential time sequence is smoothed, the media friendliness after obtaining smoothly refers to
Number time series；
Media friendliness based on target medium described in the media friendliness exponential time sequence estimation after smooth.
With reference to first aspect, in the first possible implementation, the method further includes：
Persistently obtain the multiple related text；
The target is persistently assessed with preset time window and updated to the multiple related text based on lasting acquisition
The media friendliness of media.
With reference to first aspect or the first possible realization method of first aspect, in second of possible realization method
In, described the step of obtaining multiple related texts relevant with public sentiment object that target medium is issued, further comprises：
Obtain the media text set of the target medium publication；
Generate the primary vector spatial model of each media text in the media text set；And
Generate the secondary vector spatial model of the public sentiment object；
Calculate the similarity of each the primary vector spatial model and the secondary vector spatial model；
If the similarity being calculated is more than predetermined threshold value, it is determined that the media text is the public sentiment object phase
The related text of pass.
With reference to first aspect, in the third possible realization method, each phase in the multiple related text
The step of text carries out feeling polarities analysis is closed to further comprise：
Emotional semantic classification or machine learning algorithm based on sentiment dictionary carry out feeling polarities analysis to each related text.
With reference to first aspect to the third any one possible realization method of first aspect, in the 4th kind of possible reality
In existing mode, the feeling polarities include positive emotion, neutral emotion and negative emotion, described respectively by each related text
Feeling polarities value the step of being converted to each related text corresponding media friendliness index further comprise：
According to the feeling polarities value of predefined media friendliness formula of index and each related text, count respectively
Calculation obtains the corresponding media friendliness index of each related text；
Wherein, the predefined media friendliness formula of index using it is following any one：
Fr=p+x-n；
Fs=p-x-n；
Wherein, FrFor loose media friendliness index, FsFor stringent media friendliness index, p indicates the general of positive emotion
Rate, x indicate that the probability of neutral emotion, n indicate the probability of negative emotion, 0≤p, x, n≤1, p+x+n=1, -1≤Fr,Fs≤1。
With reference to first aspect, in the 5th kind of possible realization method, the composition media friendliness exponential time sequence
The step of before, the method further includes：
Average computation is carried out according to time window to the corresponding media friendliness index of each related text, is obtained described
The mean value of the corresponding media friendliness index of time window；
The step of composition media friendliness exponential time sequence, further comprises：
Based on the corresponding multiple mean values of multiple time windows, the media friendliness exponential time sequence is constituted
Row.
With reference to first aspect or the 5th kind of possible realization method of the first or first aspect of first aspect, the 6th
It is described that the media friendliness exponential time sequence is smoothed in the possible realization method of kind, after obtaining smoothly
The step of media friendliness exponential time sequence, further comprises：
The media friendliness exponential time sequence is smoothly located using the method for moving average or the method for weighted moving average
Reason.
With reference to first aspect or the first possible realization method of first aspect, in the 7th kind of possible realization method
In, the step of the media friendliness of target medium described in the media friendliness exponential time sequence estimation based on after smooth
After rapid, the method further includes：
According to the assessment result of the media friendliness to the target medium, the media of following target medium are predicted
Friendliness index.
Second aspect, provides a kind of media friendliness apparatus for evaluating, and described device includes：
Text acquisition module, for obtain target medium publication with the relevant multiple related texts of public sentiment object；
Sentiment analysis module is obtained for carrying out feeling polarities analysis to each related text in the multiple related text
To the feeling polarities value of each related text；
Index conversion module, for the feeling polarities value of each related text to be converted to each related text respectively
Corresponding media friendliness index；
Sequence composition module, for constituting media friendliness exponential time sequence；
Smoothing module, for being smoothed to the media friendliness exponential time sequence, after obtaining smoothly
The media friendliness exponential time sequence；
Friendliness evaluation module, for based on target described in the media friendliness exponential time sequence estimation after smooth
The media friendliness of media.
In conjunction with second aspect, in the first possible implementation, the text acquisition module is specifically used for：
Persistently obtain the multiple related text；
The friendliness evaluation module is specifically used for：
The target is persistently assessed with preset time window and updated to the multiple related text based on lasting acquisition
The media friendliness of media.
In conjunction with the possible realization method of the first of second aspect or second aspect, in second of possible realization method
In, the text acquisition module is specifically used for：
Obtain the media text set of the target medium publication；
Generate the primary vector spatial model of each media text in the media text set；And
Generate the secondary vector spatial model of the public sentiment object；
Calculate the similarity of each the primary vector spatial model and the secondary vector spatial model；
If the similarity being calculated is more than predetermined threshold value, it is determined that the media text is the public sentiment object phase
The related text of pass.
In conjunction with second aspect, in the third possible realization method, the sentiment analysis module is specifically used for：
Emotional semantic classification or machine learning algorithm based on sentiment dictionary carry out feeling polarities analysis to each related text.
In conjunction with the third any one possible realization method of second aspect to second aspect, in the 4th kind of possible reality
In existing mode, the feeling polarities include that positive emotion, neutral emotion and negative emotion, the index conversion module are specifically used
In：
According to the feeling polarities value of predefined media friendliness formula of index and each related text, count respectively
Calculation obtains the corresponding media friendliness index of each related text；
Wherein, the predefined media friendliness formula of index using it is following any one：
Fr=p+x-n；
Fs=p-x-n；
Wherein, FrFor loose media friendliness index, FsFor stringent media friendliness index, p indicates the general of positive emotion
Rate, x indicate that the probability of neutral emotion, n indicate the probability of negative emotion, 0≤p, x, n≤1, p+x+n=1, -1≤Fr,Fs≤1。
In conjunction with second aspect, in the 5th kind of possible realization method, described device further includes：
Average computation block, for being carried out according to time window to the corresponding media friendliness index of each related text
Average computation obtains the mean value of the corresponding media friendliness index of the time window；
The Sequence composition module is additionally operable to be based on the corresponding multiple mean values of multiple time windows, constitutes institute
State media friendliness exponential time sequence.
In conjunction with the 5th kind of possible realization method of the first or second aspect of second aspect or second aspect, the 6th
In the possible realization method of kind, the smoothing module is specifically used for：
The media friendliness exponential time sequence is smoothly located using the method for moving average or the method for weighted moving average
Reason.
In conjunction with the possible realization method of the first of second aspect or second aspect, in the 7th kind of possible realization method
In, described device further includes：
Exponential forecasting module predicts future for the assessment result according to the media friendliness to the target medium
The media friendliness index of the target medium.
The third aspect, provides a kind of media friendliness apparatus for evaluating, and described device includes：
One or more processor；
Memory；
The program being stored in the memory, when being executed by one or more of processors, the journey
The step of sequence makes the processor execute the media friendliness appraisal procedure as described in above-mentioned first aspect any one.
Fourth aspect, provides a kind of computer readable storage medium, and the computer-readable recording medium storage has journey
Sequence, when said program is executed by a processor so that the processor executes the matchmaker as described in above-mentioned first aspect any one
The step of body friendliness appraisal procedure.
A kind of media friendliness appraisal procedure of offer of the embodiment of the present invention, device and computer readable storage medium, pass through
Based on the sentiment analysis of each related text of target medium publication, the feeling polarities value of each related text is converted to respectively
The corresponding media friendliness index of each related text, and media friendliness exponential time sequence is constituted, and to media friendliness
Exponential time sequence is smoothed, and the matchmaker of the media friendliness exponential time sequence estimation target medium after utilization smoothly
Body friendliness is achieved in the conversion from text emotion to media attitude, and target medium pair is accurately assessed so as to realize
The attitude steady in a long-term of public sentiment object.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, other are can also be obtained according to these attached drawings
Attached drawing.
Fig. 1 is a kind of media friendliness appraisal procedure flow chart provided in an embodiment of the present invention；
Fig. 2 is the result being smoothed to media friendliness exponential time sequence in the embodiment of the present invention；
Fig. 3 is a kind of block diagram of media friendliness apparatus for evaluating provided in an embodiment of the present invention.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached in the embodiment of the present invention
Figure, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only this
Invention a part of the embodiment, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art exist
The every other embodiment obtained under the premise of creative work is not made, shall fall within the protection scope of the present invention.
An embodiment of the present invention provides a kind of media friendliness appraisal procedures, with each related text of target medium publication
Based on sentiment analysis, by the way that the feeling polarities value of each related text is converted to the corresponding media of each related text respectively
Friendliness index, and media friendliness exponential time sequence is constituted, and media friendliness exponential time sequence is carried out smooth
Processing, and using the media friendliness of the media friendliness exponential time sequence estimation target medium after smooth, be achieved in from
Text emotion accurately assesses target medium to the steady in a long-term of public sentiment object to the conversion of the long-term attitude of media to realize
Attitude.Wherein, method provided in this embodiment can be applied to various equipment, such as desktop computer, personal computer, mobile terminal
And server, the present invention do not limit this.
Shown in referring to Fig.1, the embodiment of the present invention provides a kind of media friendliness appraisal procedure flow chart, and this method includes such as
Lower step：
S10, obtain target medium publication with the relevant multiple related texts of public sentiment object.
S20 carries out feeling polarities analysis to each related text in multiple related texts, obtains the emotion of each related text
Polarity number.
The feeling polarities value of each related text is converted to the corresponding media friendliness index of each related text by S30 respectively,
And constitute media friendliness exponential time sequence.
S40 is smoothed media friendliness exponential time sequence, obtain it is smooth after media friendliness index when
Between sequence.
S50, the media friendliness based on the media friendliness exponential time sequence estimation target medium after smooth.
An embodiment of the present invention provides a kind of media friendliness appraisal procedures, pass through each related text issued with target medium
Based on this sentiment analysis, it is friendly that the feeling polarities value of each related text is converted into the corresponding media of each related text respectively
Index is spent, and constitutes media friendliness exponential time sequence, and media friendliness exponential time sequence is smoothed,
And the media friendliness of the media friendliness exponential time sequence estimation target medium after utilizing smoothly, it is achieved in from text feelings
The conversion for feeling media attitude, to realize the media friendliness for accurately assessing target medium to public sentiment object.
In some embodiments of the invention, method further includes：
Persistently obtain multiple related texts；
The matchmaker of target medium is persistently assessed with preset time window and updated to multiple related texts based on lasting acquisition
Body friendliness.
Wherein, the self-defined setting of time scale that prefixed time interval can be according to the observation, for example be divided between setting time
One week, the present invention was not limited this.
Specifically, multiple related texts based on lasting acquisition, analysis obtains each in the multiple related texts for continuing to obtain
The feeling polarities value of related text, and corresponding conversion is the corresponding media friendliness index of each related text, and update was previously
The media friendliness exponential time sequence of composition, and updated media friendliness exponential time sequence is smoothed,
The media friendliness of the assessment of media friendliness exponential time sequence duration and update target medium after being then based on smoothly.
In the embodiment of the present invention, pass through the target medium publication based on lasting acquisition and the relevant multiple phases of public sentiment object
Text is closed, the media friendliness of lasting assessment and update target medium is achieved in the long-term attitude from text emotion to media
Conversion, to realize the attitude steady in a long-term for accurately assessing target medium to public sentiment object.
In some embodiments of the invention, step S10 further comprises：
Step S101 obtains the media text set of target medium publication；
Wherein, the media text set that acquisition target medium is issued in chronological order, and it is pre- to carry out text to media text set
Processing, for example Chinese word segmentation and part-of-speech tagging etc. are carried out to media text set with participle tool.
Step S102 generates the primary vector spatial model of each media text in media text set, and generates carriage
The secondary vector spatial model of feelings object.
Wherein, the process of step S102 may include：
Step S1021 is indicated media document and public sentiment object with vector respectively.
Media document DjVector can be expressed as Dj(w1j,w2j,…,wnj), wherein n is the number of words in system, wij
Keyword i is represented in document DjIn weight.
The vector of public sentiment object inquiry Q can be expressed as Q (w1k,w2k,…,wnk), wikWord i is represented in inquiring Q
Weight.
Step S1022 calculates the weight in each vector space model by the way of TF*IDF, i.e. keyword i is in document
DjWeight wij=TFij+IDFij.Specific calculating process is as follows：
TF weights：TF (Term Frequency) is the number that word occurs in a document, weight wij=TFijOr return
TF values after one change.
The normalization (Normalization) of TF：The TF values of all index terms in one document are normalized to [0,1]
Between, one of following manner usually may be used：
I), wtf=TF/max (TF)
Ii), wtf=a+ (1-a) * TF/max (TF), wherein a are regulatory factor, experience value a=0.5 or 0.4.
The document frequency DF (Document Frequency) of word：The document piece that word occurs in entire collection of document
Number, DF reflect the discrimination of word, and DF is higher, and expression word is more universal, therefore its discrimination is lower, and weight is also lower.
Inverse document frequency (Inverse DF, IDF)：The inverse of DF, (N is collection of document to the following formula calculating of generally use
In all documents number)：
Calculate weight：Keyword i is in document DjWeight wij=TFij+IDFij。
Step S103 calculates the similarity of each primary vector spatial model and secondary vector spatial model.
The degree of correlation (i.e. similarity) of document and query word can be by each vector in vectorial sky is asked opposite position
It sets to determine.There are many kinds of similarity calculation functions, and more common is two vectorial angle cosine functions.Therefore media document
DjThe similarity value that Q is inquired with public sentiment object is obtained by following formula：
Step S104, if the similarity being calculated is more than predetermined threshold value, it is determined that media text is that public sentiment object is related
Related text.
Wherein it is possible to the threshold value of inquiry correlation be set, when calculating gained similarity higher than threshold value, you can be considered phase
Document is closed, its Sentiment orientation to public sentiment object is then further analyzed.
In the embodiment of the present invention, since same media would generally issue various contents, and the analysis of public opinion is only concerned institute
The case where public sentiment object of concern, therefore the assessment of media friendliness needs selection and public sentiment visitor from the full content of media releasing
The relevant report of body, if the public sentiment object of concern can indicate that the selection of related text can be expressed as with one group of keyword
Text retrieval problem based on keyword.The present invention is used based on vector space model selection and the relevant text of public sentiment object,
The difficulty and effect of implementation process can be balanced, it is to be understood that in practical applications, selection and the relevant text of public sentiment object
Originally it can also be achieved by other methods, such as Boolean Model, vector space model, probabilistic model etc., the present invention is to obtaining
The detailed process with the relevant text of public sentiment object is taken to be not limited.
In some embodiments of the invention, step S20 further comprises：
Three polarity sentiment analysis are carried out to each related text, analysis result is respectively p, x, n, 0≤p, x, n≤1, p+x+
N=1, wherein p express the probability of positive emotion, and x indicates that the probability of neutral emotion, n indicate the probability of negative emotion.
In one embodiment, can the emotional semantic classification based on sentiment dictionary feeling polarities point are carried out to each related text
Analysis, specifically, the process may include：
Step S201 selects and improves sentiment dictionary.Wherein, sentiment dictionary includes multiple emotion words, each emotion word
Language has the score value for indicating feeling polarities intensity.
Step S202, it is that emotion word is given a mark with phrase to reinforce word and emotion negative word in conjunction with emotion.Wherein, emotion reinforces word
For the word that the degree to emotion vocabulary is modified, to indicate the intensity of emotion；Emotion negative word is to be carried out to emotion vocabulary
The word of negative, such as：, be not, be non-, absolutely not, it is no, not, can not possibly, not have, not have, fail etc. words.
Step S203 carries out cumulative statistics to the emotion score value of each polarity emotion in document, and calculates separately each pole
The probability that disposition sense occurs in a document.
In the embodiment of the present invention, three polarity emotions point are carried out to each related text by the emotional semantic classification based on sentiment dictionary
Analysis, obtains the feeling polarities value of each related text, passes through the sentiment analysis results presumption media friendliness based on single document, energy
It enough ensure that the reliability and feasibility on assessment basis.
In another embodiment, machine learning algorithm can be based on and feeling polarities analysis, tool is carried out to each related text
For body, which may include：
Step S211, construction feature：Using bag of words (n-gram), part of speech, emotion word and emotion phrase, emotion transposition
Word etc. is used as feature.
Step S212, machine learning：Using simple Bayes, maximum entropy (Maximum Entropy) and SVM as classification
Device is trained.
Step S213, each related text of combining classification device carry out feeling polarities analysis, obtain the emotion pole of each related text
Property value.
It should be noted that although in the present embodiment with to each related text carry out three polarity sentiment analysis preferably,
However, it will be understood that method provided in an embodiment of the present invention is not limited to this, those skilled in the art can also use pair
Each related text carries out two polarity sentiment analysis or multipolarity sentiment analysis.
In the embodiment of the present invention, three polarity sentiment analysis are carried out to each related text by being based on machine learning algorithm, are obtained
To the feeling polarities value of each related text, pass through the sentiment analysis results presumption media friendliness based on single document, Neng Goubao
The reliability and feasibility on assessment basis is demonstrate,proved.
In some embodiments of the invention, feeling polarities include positive emotion, neutral emotion and negative emotion, step
S30 further comprises：
Step S301, according to the feeling polarities value of predefined media friendliness formula of index and each related text,
It calculates separately to obtain the corresponding media friendliness index of each related text；
Wherein, predefined media friendliness formula of index using it is following any one：
Fr=p+x-n；
Fs=p-x-n；
Wherein, FrFor loose media friendliness index, FsFor stringent media friendliness index, p indicates the general of positive emotion
Rate, x indicate that the probability of neutral emotion, n indicate the probability of negative emotion, 0≤p, x, n≤1, p+x+n=1, -1≤Fr,Fs≤1。
Loosely difference lies in the utilization of centering disposition sense, loose friendliness disposition in stringent friendliness Index Definition
Front is regarded in sense as, and neutral emotion is regarded as negative by stringent friendliness.
Step S302 is based on the corresponding media friendliness index of each related text, when constituting a media friendliness index
Between sequence.
Wherein, a text of each element representation media releasing of media friendliness exponential time sequence is to public sentiment object
Sentiment orientation.
In the embodiment of the present invention, since the emotion that single document is embodied cannot reflect the long-term attitude of media, lead to
It crosses and the feeling polarities value of each related text is converted into the corresponding media friendliness index of each related text respectively, and constitute media
Friendliness exponential time sequence, using media friendliness exponential time sequence, to realize the long-term state from document emotion to media
The conversion of degree, and then can realize that the media friendliness to target medium for public sentiment object is assessed and tracked.
In some embodiments of the invention, step S40 further comprises：
Media friendliness exponential time sequence is smoothed using the method for moving average or the method for weighted moving average.
Wherein, the process being smoothed to media friendliness exponential time sequence using the method for moving average can wrap
It includes：
If time series is y1,y2,…,yt；The calculation formula of the simple method of moving average is：
Wherein, Mt- t phase moving averages；The item number of N- rolling averages
When n is large, calculation amount can be greatly reduced using recurrence formula.
Since rolling average does not account for the difference that different document contributes media friendliness, the method for weighted moving average can
Solve the problems, such as this.
Wherein, the process being smoothed to media friendliness exponential time sequence using weighted moving average can wrap
It includes：
Weighting sliding average general expression be：
Wherein, Mtw- t phase moving averages；wi-yt-i+1Weight, embody corresponding ytIt is important in weighted average
Property.
The selection of weight has centainly empirical, and general principle is：The flexible strategy of Recent data are big, the power of data at a specified future date
Number is small.
Wherein, weight can utilize time attenuation function f (n)=A+e-μnIt indicates, wherein A ∈ Z, and A ＞ 1, μ ∈ (0,
1), n=1,2 ..., N indicate the serial number of emotion time series.
The scene of combination public sentiment of embodiment of the present invention application, news documents are divided into original and reprint, and original document is usual
All be starting, best embody attitude of the media to public sentiment object, and forward embodied attitude tendency want it is weak very much, in order to embody
This difference, the value different from of the parameter A of the attenuation function of document weight, therefore in practical applications, can define original
Document corresponding A=1, non-original document corresponding A=0.5.Parameter μ can take random value when starting in value range, work as history
Data accumulation to a certain extent when can be estimated based on historical data, wherein fit approach can pass through python or matlab
Code is realized.
The step 40 in the embodiment of the present invention is further described in conjunction with attached drawing 2, wherein the corresponding curves of label a in Fig. 2
It is smooth preceding media friendliness exponential time sequence, the corresponding curves of label b are using the method for moving average to media friendliness
The curve that exponential time sequence obtains after being smoothed, the corresponding curves of label c are using weighted mean method to media friend
The curve obtained after exponential time sequence is smoothed is spent well, can intuitively be found out by Fig. 2, by the method for moving average and is added
The media friendliness numerical value that two methods of weight average method smoothly obtain media friendliness exponential time sequence is more steady.
In the embodiment of the present invention, by being smoothed to media friendliness exponential time sequence, after obtaining smoothly
Media friendliness exponential time sequence, and based on the media of the media friendliness exponential time sequence estimation target medium after smooth
Friendliness, thus, it is possible to ensure that the assessment result for being directed to the media friendliness for waiting for public sentiment object to target medium is more stablized, into
And it obtains target medium and is directed to the attitude steady in a long-term for waiting for public sentiment object.
In some embodiments of the invention, before constituting media friendliness exponential time sequence in step s 40, method
Can also include：
Average computation is carried out according to time window to the corresponding media friendliness index of each related text, obtains time window
The mean value of interior corresponding media friendliness index；
The step of media friendliness exponential time sequence is constituted in step S40 further comprises：
Based on the corresponding multiple mean values of multiple time windows, media friendliness exponential time sequence is constituted.
Wherein, smooth result is related with time scale, and the time scale of observation is further amplified if necessary, may be selected
A certain range of time scale, such as week, the moon, season, each related text pair in multiple related texts in this time window
The media friendliness index answered obtains the mean value of media friendliness index in time window after carrying out average computation, then just constitutes
Media friendliness exponential time sequence, and media friendliness exponential time sequence smoothly, can further analyze in this way
The media friendliness situation of big time scale.
In some embodiments of the invention, after step S50, method can also include：
According to the assessment result of the media friendliness to target medium, predict that the media friendliness of following target medium refers to
Number.
Wherein it is possible to be predicted using weighted moving average, predictor formula is：I.e. with the t phases
Predicted value of the weighted moving average as the t+1 phases.
The embodiment of the present invention is based on the sentiment analysis of multiple related texts of target medium, by analyzing single text
Corresponding media friendliness index constitutes media friendliness and indicates time series, and based on the smooth place to time series analysis
It manages to assess and predict steady in a long-term attitude of the media to public sentiment object, thus, it is possible to ensure to assess and predict media to public sentiment visitor
The accuracy of the attitude steady in a long-term of body.
Shown in please referring to Fig.3, in the embodiment of the present invention, a kind of media friendliness apparatus for evaluating is additionally provided, the device packet
It includes：
Text acquisition module 31, for obtain target medium publication with the relevant multiple related texts of public sentiment object；
Sentiment analysis module 32 is obtained for carrying out feeling polarities analysis to each related text in multiple related texts
The feeling polarities value of each related text；
It is corresponding to be converted to each related text by index conversion module 33 for respectively for the feeling polarities value of each related text
Media friendliness index；
Sequence composition module 34, for constituting media friendliness exponential time sequence；
Smoothing module 35, for being smoothed to media friendliness exponential time sequence, after obtaining smoothly
Media friendliness exponential time sequence；
Friendliness evaluation module 36, for based on the media friendliness exponential time sequence estimation target medium after smooth
Media friendliness.
Further, text acquisition module 31 is specifically used for：
Persistently obtain multiple related texts；
Friendliness evaluation module 36 is specifically used for：
The matchmaker of target medium is persistently assessed with preset time window and updated to multiple related texts based on lasting acquisition
Body friendliness.
Further, text acquisition module 31 is specifically used for：
Obtain the media text set of target medium publication；
Generate the primary vector spatial model of each media text in media text set；And
Generate the secondary vector spatial model of public sentiment object；
Calculate the similarity of each primary vector spatial model and secondary vector spatial model；
If the similarity being calculated is more than predetermined threshold value, it is determined that media text is the relevant related text of public sentiment object
This.
Further, sentiment analysis module 32 is specifically used for：
Emotional semantic classification or machine learning algorithm based on sentiment dictionary carry out feeling polarities analysis to each related text.
Further, feeling polarities include that positive emotion, neutral emotion and negative emotion, index conversion module 33 are specifically used
In：
According to the feeling polarities value of predefined media friendliness formula of index and each related text, calculate separately
To the corresponding media friendliness index of each related text；
Wherein, predefined media friendliness formula of index using it is following any one：
Fr=p+x-n；
Fs=p-x-n；
Wherein, FrFor loose media friendliness index, FsFor stringent media friendliness index, p indicates the general of positive emotion
Rate, x indicate that the probability of neutral emotion, n indicate the probability of negative emotion, 0≤p, x, n≤1, p+x+n=1, -1≤Fr,Fs≤1。
Further, device further includes：
Average computation block 37, for being put down according to time window to the corresponding media friendliness index of each related text
It calculates, obtains the mean value of corresponding media friendliness index in time window；
Sequence composition module 34 is additionally operable to be based on the corresponding multiple mean values of multiple time windows, constitutes media friendliness and refers to
Number time series.
Further, smoothing module 35 is specifically used for：
Media friendliness exponential time sequence is smoothed using the method for moving average or the method for weighted moving average.
Further, device further includes：
Exponential forecasting module 38 predicts following mesh for the assessment result according to the media friendliness to target medium
Mark the media friendliness index of media.
The embodiment of the present invention provides a kind of media friendliness apparatus for evaluating, which passes through each phase for being issued with target medium
Based on the sentiment analysis for closing text, the feeling polarities value of each related text is converted into the corresponding media of each related text respectively
Friendliness index, and media friendliness exponential time sequence is constituted, and media friendliness exponential time sequence is carried out smooth
Processing, and using the media friendliness of the media friendliness exponential time sequence estimation target medium after smooth, be achieved in from
Text emotion accurately assesses target medium to the steady in a long-term of public sentiment object to the conversion of the long-term attitude of media to realize
Attitude.
In addition, the embodiment of the present invention also provides a kind of media friendliness apparatus for evaluating, device includes：
One or more processor；
Memory；
Program stored in memory, when being executed by one or more processor, program makes processor execute
The step of stating media friendliness appraisal procedure any in embodiment.
Another embodiment of the present invention also provides a kind of computer readable storage medium, and computer-readable recording medium storage has
Program, when program is executed by processor so that processor executes media friendliness appraisal procedure any in above-described embodiment
The step of.
It should be understood by those skilled in the art that, the embodiment in the embodiment of the present invention can be provided as method, apparatus or meter
Calculation machine program product.Therefore, complete hardware embodiment, complete software embodiment can be used in the embodiment of the present invention or combine soft
The form of the embodiment of part and hardware aspect.Moreover, it wherein includes meter to be can be used in the embodiment of the present invention in one or more
The computer-usable storage medium of calculation machine usable program code (includes but not limited to magnetic disk storage, CD-ROM, optical memory
Deng) on the form of computer program product implemented.
It is with reference to the method, apparatus (system) of middle embodiment according to embodiments of the present invention and to calculate in the embodiment of the present invention
The flowchart and/or the block diagram of machine program product describes.It should be understood that can be realized by computer program instructions flow chart and/or
The combination of the flow and/or box in each flow and/or block and flowchart and/or the block diagram in block diagram.It can carry
For the processing of these computer program instructions to all-purpose computer, special purpose computer, Embedded Processor or other programmable datas
The processor of equipment is to generate a machine so that is executed by computer or the processor of other programmable data processing devices
Instruction generation refer to for realizing in one flow of flow chart or multiple flows and/or one box of block diagram or multiple boxes
The device of fixed function.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that instruction generation stored in the computer readable memory includes referring to
Enable the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one box of block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device so that count
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, in computer or
The instruction executed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Although the preferred embodiment in the embodiment of the present invention has been described, once a person skilled in the art knows
Basic creative concept, then additional changes and modifications may be made to these embodiments.So appended claims are intended to explain
It is to include preferred embodiment and fall into all change and modification of range in the embodiment of the present invention.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
God and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (18)
1. a kind of media friendliness appraisal procedure, which is characterized in that the method includes the steps：
Obtain target medium publication with the relevant multiple related texts of public sentiment object；
Feeling polarities analysis is carried out to each related text in the multiple related text, obtains the emotion of each related text
Polarity number；
The feeling polarities value of each related text is converted into the corresponding media friendliness index of each related text respectively,
And constitute media friendliness exponential time sequence；
The media friendliness exponential time sequence is smoothed, obtain it is smooth after the media friendliness index when
Between sequence；
Media friendliness based on target medium described in the media friendliness exponential time sequence estimation after smooth.
2. according to the method described in claim 1, it is characterized in that, the method further includes：
The target medium is persistently assessed with preset time window and updated to the multiple related text based on lasting acquisition
Media friendliness.
3. method according to claim 1 or 2, which is characterized in that it is described obtain target medium publication with public sentiment object
The step of relevant multiple related texts, further comprises：
Obtain the media text set of the target medium publication；
Generate the primary vector spatial model of each media text in the media text set；And
Generate the secondary vector spatial model of the public sentiment object；
Calculate the similarity of each the primary vector spatial model and the secondary vector spatial model；
If the similarity being calculated is more than predetermined threshold value, it is determined that the media text is that the public sentiment object is relevant
Related text.
4. according to the method described in claim 1, it is characterized in that, each related text in the multiple related text
The step of carrying out feeling polarities analysis further comprises：
Emotional semantic classification or machine learning algorithm based on sentiment dictionary carry out feeling polarities analysis to each related text.
5. method according to any one of claims 1 to 4, which is characterized in that the feeling polarities include positive emotion,
Neutral emotion and negative emotion, it is described that the feeling polarities value of each related text is converted into each related text pair respectively
The step of media friendliness index answered, further comprises：
According to the feeling polarities value of predefined media friendliness formula of index and each related text, calculate separately
To the corresponding media friendliness index of each related text；
Wherein, the predefined media friendliness formula of index using it is following any one：
Fr=p+x-n；
Fs=p-x-n；
Wherein, FrFor loose media friendliness index, FsFor stringent media friendliness index, p indicates the probability of positive emotion, x tables
Show that the probability of neutral emotion, n indicate the probability of negative emotion, 0≤p, x, n≤1, p+x+n=1, -1≤Fr,Fs≤1。
6. according to the method described in claim 1, it is characterized in that, the step of the composition media friendliness exponential time sequence
Before, the method further includes：
Average computation is carried out according to time window to the corresponding media friendliness index of each related text, obtains the time
The mean value of the corresponding media friendliness index of window；
The step of composition media friendliness exponential time sequence, further comprises：
Based on the corresponding multiple mean values of multiple time windows, the media friendliness exponential time sequence is constituted.
7. according to the method described in claims 1 or 2 or 5, which is characterized in that described to the media friendliness exponential time sequence
Row are smoothed, obtain it is smooth after the media friendliness exponential time sequence the step of further comprise：
The media friendliness exponential time sequence is smoothed using the method for moving average or the method for weighted moving average.
8. method according to claim 1 or 2, which is characterized in that the media friendliness based on after smooth refers to
After the step of number time series assesses the media friendliness of the target medium, the method further includes：
According to the assessment result of the media friendliness to the target medium, predict that the media of following target medium are friendly
Spend index.
9. a kind of media friendliness apparatus for evaluating, which is characterized in that described device includes：
Text acquisition module, for obtain target medium publication with the relevant multiple related texts of public sentiment object；
Sentiment analysis module obtains institute for carrying out feeling polarities analysis to each related text in the multiple related text
State the feeling polarities value of each related text；
Index conversion module is corresponded to for the feeling polarities value of each related text to be converted to each related text respectively
Media friendliness index；
Sequence composition module, for constituting media friendliness exponential time sequence；
Smoothing module, for being smoothed to the media friendliness exponential time sequence, the institute after obtaining smoothly
State media friendliness exponential time sequence；
Friendliness evaluation module, for based on target medium described in the media friendliness exponential time sequence estimation after smooth
Media friendliness.
10. device according to claim 9, which is characterized in that
The text acquisition module is specifically used for：
Persistently obtain the multiple related text；
The friendliness evaluation module is specifically used for：
The target medium is persistently assessed with preset time window and updated to the multiple related text based on lasting acquisition
Media friendliness.
11. device according to claim 9 or 10, which is characterized in that the text acquisition module is specifically used for：
Obtain the media text set of the target medium publication；
Generate the primary vector spatial model of each media text in the media text set；And
Generate the secondary vector spatial model of the public sentiment object；
Calculate the similarity of each the primary vector spatial model and the secondary vector spatial model；
If the similarity being calculated is more than predetermined threshold value, it is determined that the media text is that the public sentiment object is relevant
Related text.
12. device according to claim 9, which is characterized in that the sentiment analysis module is specifically used for：
Emotional semantic classification or machine learning algorithm based on sentiment dictionary carry out feeling polarities analysis to each related text.
13. according to the device described in claim 9 to 12 any one, which is characterized in that the feeling polarities include positive feelings
Sense, neutral emotion and negative emotion, the index conversion module are specifically used for：
According to the feeling polarities value of predefined media friendliness formula of index and each related text, calculate separately
To the corresponding media friendliness index of each related text；
Wherein, the predefined media friendliness formula of index using it is following any one：
Fr=p+x-n；
Fs=p-x-n；
Wherein, FrFor loose media friendliness index, FsFor stringent media friendliness index, p indicates the probability of positive emotion, x tables
Show that the probability of neutral emotion, n indicate the probability of negative emotion, 0≤p, x, n≤1, p+x+n=1, -1≤Fr,Fs≤1。
14. device according to claim 9, which is characterized in that described device further includes：
Average computation block, for being averaged according to time window to the corresponding media friendliness index of each related text
It calculates, obtains the mean value of the corresponding media friendliness index of the time window；
The Sequence composition module is additionally operable to be based on the corresponding multiple mean values of multiple time windows, constitutes the matchmaker
Body friendliness exponential time sequence.
15. according to the device described in claim 9 or 10 or 13, which is characterized in that the smoothing module is specifically used for：
The media friendliness exponential time sequence is smoothed using the method for moving average or the method for weighted moving average.
16. device according to claim 9, which is characterized in that described device further includes：
Exponential forecasting module predicts the described of future for the assessment result according to the media friendliness to the target medium
The media friendliness index of target medium.
17. a kind of media friendliness apparatus for evaluating, which is characterized in that described device includes：
One or more processor；
Memory；
The program being stored in the memory, when being executed by one or more of processors, described program makes
The processor executes the step of media friendliness appraisal procedure as described in any one of claim 1~8.
18. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has program, when
When described program is executed by processor so that the processor executes the media friend as described in any one of claim 1~8
The step of spending appraisal procedure well.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810330401.1A CN108595564B (en) | 2018-04-13 | 2018-04-13 | Method and device for evaluating media friendliness and computer-readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810330401.1A CN108595564B (en) | 2018-04-13 | 2018-04-13 | Method and device for evaluating media friendliness and computer-readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108595564A true CN108595564A (en) | 2018-09-28 |
| CN108595564B CN108595564B (en) | 2020-08-11 |
Family
ID=63622360
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810330401.1A Active CN108595564B (en) | 2018-04-13 | 2018-04-13 | Method and device for evaluating media friendliness and computer-readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108595564B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113282754A (en) * | 2021-06-10 | 2021-08-20 | 北京中科闻歌科技股份有限公司 | Public opinion detection method, device, equipment and storage medium for news events |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102073646A (en) * | 2009-11-23 | 2011-05-25 | 北京科技大学 | Blog group-oriented subject propensity processing method and system |
| CN102789449A (en) * | 2011-05-20 | 2012-11-21 | 日电（中国）有限公司 | Method and device for evaluating comment text |
| CN104951548A (en) * | 2015-06-24 | 2015-09-30 | 烟台中科网络技术研究所 | Method and system for calculating negative public opinion index |
| CN106951409A (en) * | 2017-03-17 | 2017-07-14 | 黄淮学院 | A kind of network social intercourse media viewpoint tendency analysis system and method |
| CN107077486A (en) * | 2014-09-02 | 2017-08-18 | 菲特尔销售工具有限公司 | Affective Evaluation system and method |
| CN107169632A (en) * | 2017-04-19 | 2017-09-15 | 广东数相智能科技有限公司 | Global media community image analysis method, device and system |
-
2018
- 2018-04-13 CN CN201810330401.1A patent/CN108595564B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102073646A (en) * | 2009-11-23 | 2011-05-25 | 北京科技大学 | Blog group-oriented subject propensity processing method and system |
| CN102789449A (en) * | 2011-05-20 | 2012-11-21 | 日电（中国）有限公司 | Method and device for evaluating comment text |
| CN107077486A (en) * | 2014-09-02 | 2017-08-18 | 菲特尔销售工具有限公司 | Affective Evaluation system and method |
| CN104951548A (en) * | 2015-06-24 | 2015-09-30 | 烟台中科网络技术研究所 | Method and system for calculating negative public opinion index |
| CN106951409A (en) * | 2017-03-17 | 2017-07-14 | 黄淮学院 | A kind of network social intercourse media viewpoint tendency analysis system and method |
| CN107169632A (en) * | 2017-04-19 | 2017-09-15 | 广东数相智能科技有限公司 | Global media community image analysis method, device and system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113282754A (en) * | 2021-06-10 | 2021-08-20 | 北京中科闻歌科技股份有限公司 | Public opinion detection method, device, equipment and storage medium for news events |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108595564B (en) | 2020-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hu et al. | Few-shot charge prediction with discriminative legal attributes | |
| CA3129745C (en) | Neural network system for text classification | |
| US20210319051A1 (en) | Conversation oriented machine-user interaction | |
| Bagheri et al. | ADM-LDA: An aspect detection model based on topic modelling using the structure of review sentences | |
| CN112560479B (en) | Abstract extraction model training method, abstract extraction device and electronic equipment | |
| CN109657054A (en) | Abstraction generating method, device, server and storage medium | |
| Wu et al. | Controllable abstractive dialogue summarization with sketch supervision | |
| Paetzold et al. | Sv000gg at semeval-2016 task 11: Heavy gauge complex word identification with system voting | |
| US10067935B2 (en) | Prediction and optimized prevention of bullying and other counterproductive interactions in live and virtual meeting contexts | |
| US11032217B2 (en) | Reusing entities in automated task-based multi-round conversation | |
| KR20110132991A (en) | Identifying activities using a hybrid user-activity model | |
| CN110971659A (en) | Recommendation message pushing method and device and storage medium | |
| CN108228808B (en) | Method and device for determining hot event, storage medium and electronic equipment | |
| EP4134900A2 (en) | Method and apparatus for recommending content, method and apparatus for training ranking model, device, and storage medium | |
| CN110895656B (en) | Text similarity calculation method and device, electronic equipment and storage medium | |
| CN110187780A (en) | Long text prediction technique, device, equipment and storage medium | |
| Ma et al. | Implicit discourse relation identification for open-domain dialogues | |
| Sun et al. | Gaussian word embedding with a wasserstein distance loss | |
| Sukumar et al. | Semantic based sentence ordering approach for multi-document summarization | |
| CN109271624A (en) | A kind of target word determines method, apparatus and storage medium | |
| Parker et al. | Named entity recognition through deep representation learning and weak supervision | |
| CN108595564A (en) | Media friendliness appraisal procedure, device and computer readable storage medium | |
| Silva et al. | Evaluating Pre-training Strategies for Literary Named Entity Recognition in Portuguese | |
| Zhu et al. | Order-sensitive keywords based response generation in open-domain conversational systems | |
| CN114416941A (en) | Generation method and device of dialogue knowledge point determination model fusing knowledge graph |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240306 Address after: Room 1179, W Zone, 11th Floor, Building 1, No. 158 Shuanglian Road, Qingpu District, Shanghai, 201702 Patentee after: Shanghai Zhongan Information Technology Service Co.,Ltd. Country or region after: China Address before: 518000 Room 201, building A, No. 1, Qian Wan Road, Qianhai Shenzhen Hong Kong cooperation zone, Shenzhen, Guangdong (Shenzhen Qianhai business secretary Co., Ltd.) Patentee before: ZHONGAN INFORMATION TECHNOLOGY SERVICE Co.,Ltd. Country or region before: China |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240415 Address after: Room 1179, W Zone, 11th Floor, Building 1, No. 158 Shuanglian Road, Qingpu District, Shanghai, 201702 Patentee after: Shanghai Zhongan Information Technology Service Co.,Ltd. Country or region after: China Address before: 518000 Room 201, building A, No. 1, Qian Wan Road, Qianhai Shenzhen Hong Kong cooperation zone, Shenzhen, Guangdong (Shenzhen Qianhai business secretary Co., Ltd.) Patentee before: ZHONGAN INFORMATION TECHNOLOGY SERVICE Co.,Ltd. Country or region before: China |
|
| TR01 | Transfer of patent right |