CN110516249A - A kind of Sentiment orientation information obtaining method and device - Google Patents
A kind of Sentiment orientation information obtaining method and device Download PDFInfo
- Publication number
- CN110516249A CN110516249A CN201910807068.3A CN201910807068A CN110516249A CN 110516249 A CN110516249 A CN 110516249A CN 201910807068 A CN201910807068 A CN 201910807068A CN 110516249 A CN110516249 A CN 110516249A
- Authority
- CN
- China
- Prior art keywords
- word
- information
- processed
- sentiment orientation
- comment information
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000012545 processing Methods 0.000 claims description 81
- 230000008451 emotion Effects 0.000 claims description 51
- 238000012512 characterization method Methods 0.000 claims description 32
- 238000013136 deep learning model Methods 0.000 claims description 19
- 238000003860 storage Methods 0.000 claims description 16
- 230000011218 segmentation Effects 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 8
- 101100517651 Caenorhabditis elegans num-1 gene Proteins 0.000 claims description 4
- 235000013399 edible fruits Nutrition 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 12
- 230000036651 mood Effects 0.000 description 12
- 238000004458 analytical method Methods 0.000 description 5
- 230000033228 biological regulation Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000003796 beauty Effects 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000005498 polishing Methods 0.000 description 1
- 230000008929 regeneration Effects 0.000 description 1
- 238000011069 regeneration method Methods 0.000 description 1
Abstract
The embodiment of the present application provides a kind of Sentiment orientation information obtaining method and device, is related to Internet technical field, wherein the above method includes: to obtain comment information to be processed, and the comment information to be processed is the comment information of information to be analyzed；Calculate the importance degree of word that the comment information to be processed includes in the comment information to be processed；According to the sequence of the importance degree being calculated from high to low, from the word that the comment information to be processed includes, selection has representational word to the comment information to be processed；According to selected word, the Sentiment orientation information of the information to be analyzed is obtained.Using scheme provided by the embodiments of the present application, the Sentiment orientation information of information can be obtained.
Description
Technical field
This application involves Internet technical fields, more particularly to a kind of Sentiment orientation information obtaining method and device.
Background technique
With the fast development of Internet technology, the information that internet can provide for user is more and more extensive.And it is above-mentioned
Various information often have Sentiment orientation.For example, the Sentiment orientation of an information, which can be, indicates that the emotion of positive mood is inclined
To, indicate the Sentiment orientations etc. of negative emotions.Since the information with negative emotions may bring bad shadow for users
It rings, for this reason, it may be necessary to the Sentiment orientation information of information is obtained, the various information that then internet is provided according to Sentiment orientation information
It is monitored.
In view of the foregoing, it is desirable to provide a kind of scheme for the Sentiment orientation information for obtaining information.
Summary of the invention
The embodiment of the present application is designed to provide a kind of Sentiment orientation information obtaining method and device, to obtain information
Sentiment orientation information.Specific technical solution is as follows:
In a first aspect, the embodiment of the present application provides a kind of Sentiment orientation information obtaining method, which comprises
Comment information to be processed is obtained, the comment information to be processed is the comment information of information to be analyzed；
Calculate the importance degree of word that the comment information to be processed includes in the comment information to be processed；
According to the sequence of the importance degree being calculated from high to low, the word that includes from the comment information to be processed
In, selection has representational word to the comment information to be processed；
According to selected word, the Sentiment orientation information of the information to be analyzed is obtained.
Second aspect, the embodiment of the present application provide a kind of Sentiment orientation information obtaining device, and described device includes:
Comment information obtains module, and for obtaining comment information to be processed, the comment information to be processed is letter to be analyzed
The comment information of breath；
Importance degree computing module, for calculating word that the comment information to be processed includes in the comment to be processed
Importance degree in information；
Word selecting module to be processed is commented from described for the sequence according to the importance degree being calculated from high to low
In the word for including by information, selection has representational word to the comment information to be processed；
Sentiment orientation information acquisition module, for obtaining the Sentiment orientation of the information to be analyzed according to selected word
Information.
The third aspect, the embodiment of the present application provide a kind of electronic equipment, including processor and machine readable storage medium,
The machine readable storage medium is stored with the machine-executable instruction that can be executed by the processor, and the processor is by institute
It states machine-executable instruction to promote: realizing Sentiment orientation information obtaining method step described in the embodiment of the present application.
Fourth aspect, the embodiment of the present application provide a kind of machine readable storage medium, are stored with machine-executable instruction,
When being called and being executed by processor, the machine-executable instruction promotes the processor: realizing described in the embodiment of the present application
Sentiment orientation information obtaining method step.
As seen from the above, when obtaining the Sentiment orientation information of information to be analyzed using scheme provided by the embodiments of the present application,
It is the comment information based on information to be analyzed, that is, comment information to be processed obtains.Include according to comment information to be processed
Importance degree of the word in comment information to be processed, from the word that comment information to be processed includes, selection is commented to be processed
By information there is representational word to obtain the Sentiment orientation information of information to be analyzed then according to selected word.Due to wait locate
Reason comment information reflects Sentiment orientation of the reader to information to be analyzed of information to be analyzed, and due to above-mentioned selected word
To comment information to be processed have it is representational, so, the reader that above-mentioned selected word is able to reflect information to be analyzed treats
Analyze the Sentiment orientation of information.Therefore, the Sentiment orientation information of information to be analyzed can be obtained according to above-mentioned selected word.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of application for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow diagram of the first Sentiment orientation information obtaining method provided by the embodiments of the present application；
Fig. 2 is the flow diagram of second of Sentiment orientation information obtaining method provided by the embodiments of the present application；
Fig. 3 is the flow diagram of the third Sentiment orientation information obtaining method provided by the embodiments of the present application；
Fig. 4 is the structural schematic diagram of the first Sentiment orientation information obtaining device provided by the embodiments of the present application；
Fig. 5 is the structural schematic diagram of second of Sentiment orientation information obtaining device provided by the embodiments of the present application；
Fig. 6 is the structural schematic diagram of the third Sentiment orientation information obtaining device provided by the embodiments of the present application；
Fig. 7 is the structural schematic diagram of a kind of electronic equipment provided by the embodiments of the present application.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.
Since the information that internet provides a user is more and more extensive, and above-mentioned various information are often inclined with emotion
To need to obtain the Sentiment orientation information of information convenient for being monitored to the various information that internet provides.Based on this, originally
Application embodiment provides a kind of Sentiment orientation information obtaining method and device.
In one embodiment of the application, a kind of Sentiment orientation information obtaining method is provided, this method comprises:
Comment information to be processed is obtained, comment information to be processed is the comment information of information to be analyzed；
Calculate the importance degree of word that comment information to be processed includes in comment information to be processed；
According to the sequence of the importance degree being calculated from high to low, from the word that comment information to be processed includes, choosing
Selecting has representational word to comment information to be processed；
According to selected word, the Sentiment orientation information of information to be analyzed is obtained.
Since above-mentioned comment information to be processed reflects Sentiment orientation of the reader to information to be analyzed of information to be analyzed,
Again due to above-mentioned selected word to comment information to be processed have it is representational, so, above-mentioned selected word be able to reflect to
Analyze Sentiment orientation of the reader to information to be analyzed of information.Therefore, it can be obtained according to above-mentioned selected word to be analyzed
The Sentiment orientation information of information.
Specific embodiment is first passed through below to carry out in detail Sentiment orientation information obtaining method provided by the embodiments of the present application
Explanation.
Referring to Fig. 1, the flow diagram of the first Sentiment orientation information obtaining method is provided, this method includes following step
Rapid S101-S104.
S101: comment information to be processed is obtained.
Wherein, above-mentioned comment information to be processed is the comment information of information to be analyzed.
Specifically, above-mentioned information to be analyzed can be the information such as news, the network novel, forum's topic, video, audio.
For information to be analyzed for one, comment information can refer to all comment informations of the information to be analyzed,
It can also refer to the part comment information of the information to be analyzed.For example, newest first in the comment information of the information to be analyzed
The comment information delivered in nearest preset duration in preset quantity comment information, the comment information of the information to be analyzed, should be to
Analyze the comment information etc. that length in the comment information of information reaches preset length.
S102: the importance degree of word that comment information to be processed includes in comment information to be processed is calculated.
The word that comment information to be processed includes used language when making comments information by the reader of information to be analyzed
Speech determines.For example, in the case that the language used when above-mentioned reader makes comments information is Chinese, above-mentioned comment to be processed
The word that information includes can be monosyllabic word, the multi-character words etc. in Chinese, e.g., good, news, beauty, joy etc..In above-mentioned reading
Person make comments information when the language that uses for English in the case where, the word that above-mentioned comment information to be processed includes can be English
In word, e.g., news, happy, apple etc..In view of the foregoing, the word that above-mentioned comment information to be processed includes can be with
For day cliction, method cliction etc., the embodiment of the present application is defined not to this.It is to be processed in one embodiment of the application
The word for including in comment information can be obtained by word segmentation processing.Specifically, can be calculated using word segmentation processing in the prior art
Method realizes above-mentioned word segmentation processing, and I will not elaborate.
In addition, when calculating the importance degree of word that comment information to be processed includes in comment information to be processed, it can
It is calculated at least one of information such as the number, the part of speech that occur in comment information to be processed according to each word.It calculates
The concrete mode of above-mentioned importance degree may refer to be illustrated in fig. 2 shown below embodiment, wouldn't be described in detail here.
For a word, above-mentioned part of speech refers to the grammatical attribute of a word, is foundation word in combination or sentence
What grammatical function determined.For example, adjective, noun etc..Part of speech is the foundation sorted out to a word.
S103: according to the sequence of the importance degree being calculated from high to low, the word that includes from comment information to be processed
In, selection has representational word to comment information to be processed.
The importance degree of word is higher in comment information to be processed, illustrates that the word is more important in comment information to be processed,
To the representational stronger of comment information to be processed, for this purpose, the sequence according to the above-mentioned importance degree being calculated from high to low,
Can be from the word that comment information to be processed includes, selecting has representational word to comment information to be processed.
Specifically, can be according to the sequence of the importance degree being calculated from high to low, from comment information packet to be processed
The second preset quantity word is selected in the word contained, and selected word is representational as having to comment information to be processed
Word.
For example, above-mentioned second preset quantity can be 90,100,110 etc..
In addition, the quantity of comment information is different, and then above-mentioned comment to be processed for different information to be analyzed
The quantity for the word for including in information is also indefinite.Therefore, the quantity of word included in above-mentioned comment information to be processed may be big
In, be equal to above-mentioned second preset quantity, it is also possible to be less than above-mentioned second preset quantity.
It, can be according to important when the quantity of above-mentioned the included word of comment information to be processed is greater than above-mentioned second preset quantity
The sequence of property degree from high to low, from the word that comment information to be processed is included, select the second preset quantity word as pair
Comment information to be processed has representational word.
Above-mentioned the included word of comment information to be processed quantity be equal to above-mentioned second preset quantity when, can by it is above-mentioned to
All words that processing comment information includes have representational word as to comment information to be processed.
Above-mentioned the included word of comment information to be processed quantity be less than above-mentioned second preset quantity when, can by it is above-mentioned to
All words that processing comment information includes have representational word as to comment information to be processed.Furthermore it is possible to be supplemented one
A little preset function words have representational word as to comment information to be processed, so that having to comment information to be processed representational
Word quantity be equal to above-mentioned second preset quantity.It can guarantee to be wrapped even if in the comment information of different information to be analyzed in this way
Quantity containing word is different, and also can still select identical quantity has representational word, so that subsequent acquisition is each wait divide
When analysing the Sentiment orientation information of information, the data volume for being based on data is consistent.Certainly, included in above-mentioned comment information to be processed
When the quantity of word is less than above-mentioned second preset quantity, above-mentioned function word can not also be supplemented herein.
Wherein, above-mentioned function word is not it is to be understood that have substantive representational word to above-mentioned comment information to be processed.Example
Such as, " ", " ground " etc..
In one embodiment of the application, it is less than third present count in the quantity of above-mentioned the included word of comment information to be processed
When amount, it is believed that the word for including in comment information to be processed is very few, be not enough to for the Sentiment orientation to information to be analyzed into
Row analysis can terminate the process for obtaining Sentiment orientation information in this case.
For example, above-mentioned third preset quantity can be 120,110,100,90,60,50 etc..
Above-mentioned third preset quantity is not directly dependent upon in value with above-mentioned second preset quantity, third preset quantity and
The value of second preset quantity can be preset according to concrete application scene.
S104: according to selected word, the Sentiment orientation information of information to be analyzed is obtained.
Wherein, above-mentioned Sentiment orientation information can be the word for indicating emotion, can be the word for indicating emotion
Mark can also be expression from positive mood to the number etc. of negative sense mood grade.
For example, the word of above-mentioned expression emotion may is that it is excited, excited, glad, tranquil, unhappy, sad etc..
The mark of the word of above-mentioned expression emotion can be as follows respectively:
It is excited: 1, excited: 2, glad: 3, tranquil: 4, unhappy: 5, it is sad: 6.
It is above-mentioned that five grades can be divided into from positive mood to negative sense mood, such as:
Strong positive mood is indicated with 1, is first grade；
A point forward mood is indicated with 2, is second grade；
The neutral mood neither forward direction nor negative sense is indicated with 3, is three grades；
Some negative sense moods are indicated with 4, are the 4th grade；
Strong negative sense mood is indicated with 5, is the 5th grade.
In one embodiment of the present of invention, the emotion of selected word can be obtained, and counts various emotions
Word quantity, then divide above-mentioned grade according to the obtained quantity of statistics.
Furthermore it is also possible to calculate the word of various emotions in selected word according to the quantity that above-mentioned statistics obtains
Shared ratio, then according to the above-mentioned grade of ratio cut partition being calculated.
For example, if emotion is greater than the first preset ratio for the word of " excitement " ratio shared in selected word,
The Sentiment orientation information for then dividing above-mentioned information to be analyzed is first grade；
If emotion is the word of " happiness ", ratio shared in selected word is greater than the second preset ratio, divides
The Sentiment orientation information of above-mentioned information to be analyzed is second grade；
If emotion is the word of " calmness ", ratio shared in selected word is greater than third preset ratio, divides
The Sentiment orientation information of above-mentioned information to be analyzed is three grades；
If emotion is the word of " unhappy ", ratio shared in selected word is greater than the 4th preset ratio, draws
The Sentiment orientation information for dividing above-mentioned information to be analyzed is the 4th grade；
If emotion is the word of " sadness ", ratio shared in selected word is greater than the 5th preset ratio, divides
The Sentiment orientation information of above-mentioned information to be analyzed is the 5th grade.
Such as, above-mentioned first preset ratio to the 5th preset ratio can be respectively 80%, 70%, 60%, 70%, 80%
Deng.
In one embodiment of the present of invention, the emotion of selected word can be determined, then, for identified every
One emotion counts the quantity for belonging to the word of the emotion in selected word, according to the obtained quantity of statistics, obtain to
Analyze the Sentiment orientation information of information.
Specifically, the corresponding relationship of various words and emotion can be preset, then by inquiring above-mentioned corresponding pass
System determines the emotion of selected word.
In addition, the Sentiment orientation due to information to be analyzed can be indicated by an information, can also be indicated by how each information,
Therefore, the quantity obtained according to statistics when obtaining the Sentiment orientation information of information to be analyzed, can will count obtained quantity
In, Sentiment orientation information of the corresponding emotion of the 4th preset quantity maximum quantity as information to be analyzed.
Above-mentioned 4th preset quantity can be 1,2 etc..
It is assumed that above-mentioned 4th preset quantity is 1, the word for belonging to " happiness " this emotion in selected each word has
50, and the word for belonging to " sadness " this emotion has 5, does not belong to the word of other emotions, then can will to point
The Sentiment orientation information of analysis information is determined as " happiness ".
It is assumed that above-mentioned 4th preset quantity is 2, the word for belonging to " happiness " this emotion in selected each word has
50, the word for belonging to " calmness " this emotion has 30, and the word for belonging to " sadness " this emotion has 5, does not have
Belong to the word of other emotions, then the Sentiment orientation information that can be analysed to information is determined as " happiness " and " calmness ".
The Sentiment orientation information for obtaining information to be analyzed can also obtain by other means, wouldn't be described in detail, be detailed in here
Subsequent embodiment illustrated in fig. 3.
It as seen from the above, is base when obtaining the Sentiment orientation information of information to be analyzed using scheme provided in this embodiment
In the comment information of information to be analyzed, that is, comment information to be processed obtains.The word for including according to comment information to be processed
Importance degree in comment information to be processed, from the word that comment information to be processed includes, comment to be processed is believed in selection
There is breath representational word to obtain the Sentiment orientation information of information to be analyzed then according to selected word.It is commented due to be processed
By the reader of message reflection information to be analyzed to the Sentiment orientation of information to be analyzed, and above-mentioned selected word is to be processed
Comment information have it is representational, so, above-mentioned selected word is able to reflect the reader of information to be analyzed to information to be analyzed
Sentiment orientation.Therefore, the Sentiment orientation information of information to be analyzed can be obtained according to above-mentioned selected word.
In one embodiment of the application, referring to fig. 2, the process for providing second of Sentiment orientation information obtaining method is shown
It is intended to, compared with above-mentioned embodiment illustrated in fig. 1, in the present embodiment, above-mentioned S102 calculates the word that comment information to be processed includes and exists
Importance degree in comment information to be processed, includes the following steps S102A-S102C.
S102A: word segmentation processing is carried out to comment information to be processed, obtains the first processing result.
Specifically, can realize above-mentioned word segmentation processing using word segmentation processing algorithm in the prior art, I will not elaborate.
S102B: the stop words in the first processing result of removal obtains second processing result.
Wherein, above-mentioned stop words are as follows: preset word without emotion and/or do not have table to content of text
The word of sign property.
For example, it is that the above-mentioned word without emotion may is that,, uh etc., it is above-mentioned not have to content of text
Characterize new word may is that you, I, he, it etc..
It is difficult to reflect letter to be analyzed due to not having the word of emotion and not having representational word to content of text
Therefore the Sentiment orientation of breath is largely tool in obtained second processing result after removing upper predicate in the first processing result
The word of passionate color has representational word to content of text.For example, it is beautiful, happy, praise, be sad, is sad, more, forever
Deng.Be conducive to the subsequent Sentiment orientation information for accurately obtaining information to be analyzed in this way.
S102C: importance degree of each word in comment information to be processed in second processing result is calculated.
Since above-mentioned first processing result is to carry out word segmentation processing to comment information to be processed to obtain, so the first processing
As a result comprising each word for including in comment information to be processed in, then second processing is the result is that in the first processing result of removal
What stop words obtained, so also can include the word for including in second processing result in comment information to be processed.
In one embodiment of the application, can by following steps A-C calculate second processing result in each word to
Handle the importance degree in comment information.
Step A: the number that the sum and each word of word included in statistics second processing result occur.
Wherein, word included in second processing result is it is to be understood that all words for including in second processing result, these
There may be dittographs in word, it is also possible to dittograph be not present.Based on this, the sum of word included in second processing result
It also is just the total quantity for all words for including in second processing result.
For example, word included in second processing result can be with are as follows: glad, laugh, is tranquil, is glad, is happy, is glad.This
In the case of kind, the sum of word included in above-mentioned second processing result are as follows: 6.
Step B: for each word in second processing result, the number of the information in preset information bank comprising the word is obtained
It measures, and calculates the first characterization value and the second characterization value of the word according to following formula:
Wherein, S1 indicates that the first characterization value, S2 indicate that the second characterization value, Num1 indicate that the word occurs in second processing result
Number, Num2 indicates the sum of word included in second processing result, the sum of information included in Num3 information bank, Num4
Indicate the quantity of the information in information bank comprising the word.
Above- mentioned information library is for storing the various types of information collected in advance.The information that these are collected in advance is by specific
Application scenarios determine.
It is assumed that amounting in above-mentioned second processing result includes 100 words, wherein word X occurs in the second processing result
40 times, then above-mentioned S1=40/100=0.4.
It is assumed that it includes 1000 information that above-mentioned preset information bank is total, wherein the information comprising word X is 199 total, then
Above-mentioned S2=log (1000/ (199+1))=log (5)=0.699.
Step C: according to the first characterization value and the second characterization value of word each in second processing result, each word is calculated separately
Importance degree in comment information to be processed.
It, can be by the first characterization value for each word in second processing result in one embodiment of the application
Importance degree with the product of the second characterization value as the word in comment information to be processed.
For example, for the citing in above-mentioned steps B, importance degree of the upper predicate X in comment information to be processed are as follows:
0.4x0.699=0.2796.
In another embodiment of the application, when calculating importance degree of each word in comment information to be processed,
Other than considering above-mentioned first characterization value and the second characterization value, the part of speech of word can also be considered further that.For example, being different parts of speech
Different regulation coefficients is set, and the result that then above-mentioned product regulation coefficient corresponding with the part of speech of word is multiplied is as above-mentioned heavy
The property wanted degree, or the result that above-mentioned product regulation coefficient corresponding with the part of speech of word is added is as above-mentioned importance degree
Deng.
For example, it is assumed that regulation coefficient is 1.1 when part of speech is adjective, regulation coefficient is 0.8 when part of speech is noun.
Then for " happiness " this adjective, after the product M of the first characterization value and the second characterization value is calculated,
Using the result of M*1.1 as the importance degree of " happiness " this word, or using the result of M+1.1 as " happiness " this word
Importance degree.
For " Xiao Ming " this noun, after the product N of the first characterization value and the second characterization value is calculated, by N*
Importance degree of 0.8 result as " Xiao Ming " this word, or using the result of N+0.8 as the important of " Xiao Ming " this word
Property degree.
In one embodiment of the application, on the basis of above-mentioned steps A- step C, above-mentioned S103 selection is commented to be processed
When there is representational word by information, the also sequence according to the importance degree being calculated from high to low accordingly, from second
In the word that processing result includes, selection has representational word to comment information to be processed.
Specifically, when the quantity of above-mentioned the included word of second processing result is greater than above-mentioned second preset quantity, it can be by
According to the sequence of importance degree from high to low, the second preset quantity word is selected to comment from second processing result as to be processed
There is representational word by information.
It, can be by above-mentioned second when the quantity of above-mentioned the included word of second processing result is equal to above-mentioned second preset quantity
All words that processing result includes have representational word as to comment information to be processed.
It, can be by above-mentioned second when the quantity of above-mentioned the included word of second processing result is less than above-mentioned second preset quantity
All words that processing result includes have representational word as to comment information to be processed, furthermore it is possible to be supplemented some pre-
If function word as to comment information to be processed have representational word so as to comment information to be processed have representational word
Reach above-mentioned second preset quantity.
In addition, in one embodiment of the application, it is less than above-mentioned the in the quantity of above-mentioned the included word of second processing result
When three preset quantities, it is believed that the word for including in second processing result is very few, is not enough to for the emotion to information to be analyzed
Tendency is analyzed, and in this case, can terminate the process for obtaining Sentiment orientation information.
As seen from the above, when obtaining the Sentiment orientation information of information to be analyzed using scheme provided in this embodiment, due to
The stop words in comment information to be processed is eliminated, can to select has representational word to comment information to be processed
It is representational stronger, also more acurrate, so that the Sentiment orientation information of the information to be analyzed obtained is more accurate.
In one embodiment of the application, it can also be inclined by means of deep learning model, the emotion for obtaining information to be analyzed
To information.In this case, it needs to carry out model training in advance, obtains the model of the Sentiment orientation information for obtaining information.
Specifically, can be trained by following steps D-G to preset deep learning model, obtain for being believed
The model of the Sentiment orientation information of breath.
Step D: obtaining in the word that the comment information of sample information is included has representational sample word to sample information.
In a kind of situation, above-mentioned sample word can be staff's hand from the word that the comment information of sample information is included
Dynamic selection.
In another case, above-mentioned sample word can also be and obtain in the following manner:
Word segmentation processing is carried out to the comment information of sample information, obtains word segmentation result；Remove the stop words in word segmentation result；
Calculate importance degree of the remaining each word in the comment information of sample information in above-mentioned word segmentation result；According to being calculated
Importance degree, from word remaining in above-mentioned word segmentation result, selection to sample information have representational word, as sample
This word.This process sees the aforementioned treatment process to comment information to be processed, and details are not described herein.
Step E: encoding each sample word respectively, obtains each sample word and is compiled with the sample that binary numeral indicates
Code result.
Specifically, can be carried out according to preset encoding relation to each sample word when being encoded to each sample word
Coding.Wherein, word can be converted to the encoding relation of binary numeral by above-mentioned encoding relation for any one, and the application is implemented
Example is not defined the concrete form of above-mentioned encoding relation.Such as, above-mentioned coding result may is that 010101 etc..
For example, above-mentioned encoding relation can with the corresponding relationship of word as defined in Hanzi code for interchange and binary numeral, it is,
The corresponding relationship of word and binary numeral as defined in GB code can also be that word as defined in internal code is corresponding with binary numeral
Relationship etc..
Step F: according to above-mentioned encoding samples as a result, generating the corresponding vector of each sample word.
Wherein, the corresponding vector of each sample word includes preset quantity binary numeral.That is, each sample word
It include preset quantity element in corresponding vector, and the value of each element is a binary numeral.
Above-mentioned preset quantity can be 200,300 etc..
Each sample word generates the vector comprising preset quantity element, it is ensured that vector length phase generated
Deng.
Specifically, the quantity of binary numeral may in obtained coding result for each sample word
It can be in above-mentioned coding result to guarantee that the vector ultimately generated includes preset quantity element less than above-mentioned preset quantity
Increase Plus "0" or " 1 " before or after the binary numeral for including.
Step G: inputting preset deep learning model for the corresponding vector of each sample word and be trained, and obtains emotion and inclines
To acquisition model.
Trained detailed process is as follows step G1-G3.
Step G1: the mark Sentiment orientation information of above-mentioned sample information is obtained.
The mark Sentiment orientation information of above-mentioned sample information can be what staff marked manually.For example, above-mentioned mark
Sentiment orientation information can be expression from positive mood to the number of negative sense mood grade.Such as, above-mentioned mark Sentiment orientation information
It can be 1,2,3,4,5 etc..In this case, it is also assumed that it is a disaggregated model that above-mentioned Sentiment orientation, which obtains model,.
Step G2: the corresponding vector of each sample word is inputted into preset deep learning model, to above-mentioned sample information
Sentiment orientation is predicted, prediction Sentiment orientation information is obtained.
Step G3: calculating mark Sentiment orientation information and predicts the loss between Sentiment orientation information.
For example, mark Sentiment orientation information can be calculated and predict the difference between Sentiment orientation information, and by above-mentioned difference
Value is used as above-mentioned loss.
Step G4: it is adjusted according to model parameter of the above-mentioned loss to above-mentioned deep learning model, if after adjusting parameter
Deep learning model be unsatisfactory for preset model training and terminate to require, then repeat step G1-G3 and according to above-mentioned damage
The step of mistake is adjusted the model parameter of above-mentioned deep learning model, until meeting above-mentioned model training terminates to require, and
Model is obtained using the deep learning model after adjusting parameter as Sentiment orientation.
The specific network structure of above-mentioned deep learning model is not limited in the embodiment of the present application, it is any to be based on deep learning
Network model can be used as above-mentioned preset deep learning model.
For example, above-mentioned preset deep learning model can be based on convolutional neural networks and/or full Connection Neural Network
Model.
Specifically, when the parameter to above-mentioned deep learning model is adjusted, adjustment direction are as follows: in deep learning model
Input information be the corresponding vector of above-mentioned sample word in the case where, the Sentiment orientation information of output is to above-mentioned mark Sentiment orientation
The direction that information is drawn close.
For example, it is assumed that above-mentioned mark Sentiment orientation information are as follows: indicate the mark 1 of the word " excitement " of emotion, and it is above-mentioned
Predict Sentiment orientation information are as follows: indicate the mark 3 of the word " calmness " of emotion.Then above-mentioned adjustment direction are as follows: in deep learning
In the case that the input of model is constant, so that the direction that trend information is drawn close from " 3 " to " 1 " the case where output.
Specifically, the number that above-mentioned model training terminates to require can be execution above-mentioned steps G2-G4 reaches preset time
Number, above-mentioned model training terminate to require can also be the prediction Sentiment orientation information and mark feelings of above-mentioned deep learning model output
The loss felt between trend information is less than preset loss threshold value.The application is only illustrated as example, not to above-mentioned
Model training terminates to require to be defined.
On the basis of obtaining above-mentioned Sentiment orientation acquisition model, referring to Fig. 3, the third Sentiment orientation information is provided
The flow diagram of preparation method, compared with above-mentioned embodiment illustrated in fig. 1, in the present embodiment, above-mentioned S104 is according to selected
Word obtains the Sentiment orientation information of information to be analyzed, includes the following steps S104A-S104C.
S104A: encoding selected each word respectively, and obtain selected each word is indicated with binary numeral
Coding result.
When being encoded to selected each word, it can be realized using afore-mentioned code algorithm.Coding referred to herein
Algorithm is identical as encryption algorithm involved in abovementioned steps E, and which is not described herein again.
S104B: according to above-mentioned coding result, the corresponding vector of selected each word is generated.
Wherein, the corresponding vector of each selected word includes above-mentioned preset quantity binary numeral.
Selected each word generates the vector comprising preset quantity element, can guarantee so obtained each
Vector length is equal.
Specifically, for each selected word, after being encoded to it, include in gained coding result two
Binary value number may be less than above-mentioned preset quantity, in this case, the binary number that can include in coding result
Increase Plus "0" or " 1 " polishing before or after value to above-mentioned preset quantity.
For most of deep learning models, it may require that the data volume of its input information is identical in application process,
It,, can be with if the quantity of above-mentioned selected word is less than aforementioned second preset quantity in one embodiment of the application based on this
After generating the corresponding above-mentioned vector of selected each word, the 4th quantity preset quantity of regeneration dimension, element value are 0
Vector, and the vector that the corresponding vector sum element value generated of above-mentioned selected each word is 0 is together as above-mentioned feelings
Sense tendency obtains the input information of model.
Wherein, above-mentioned 4th quantity=aforementioned second preset quantity-selected word quantity.
S104C: by the corresponding vector of selected each word be input to the Sentiment orientation obtain model to it is described to point
The Sentiment orientation of analysis information is predicted, the Sentiment orientation information of the information to be analyzed is obtained.
As seen from the above, it when obtaining the Sentiment orientation information of information to be analyzed using scheme provided in this embodiment, is selecting
It selects after there is representational word to comment information to be processed, obtains model by means of Sentiment orientation trained in advance, obtain wait divide
Analyse the Sentiment orientation information of information.Since there are great amount of samples information, deep learning model can be very good to learn
The rule that the Sentiment orientation of sample information is shown is practised out, for this purpose, obtaining model by trained Sentiment orientation can be quasi-
The Sentiment orientation information of true acquisition information to be analyzed.
Corresponding with above-mentioned Sentiment orientation information obtaining method, the embodiment of the present application also provides a kind of Sentiment orientation information
Obtain device.
Referring to fig. 4, the structural schematic diagram of the first Sentiment orientation information obtaining device is provided, which includes:
Comment information obtains module 401, and for obtaining comment information to be processed, the comment information to be processed is to be analyzed
The comment information of information；
Importance degree computing module 402, for calculating word that the comment information to be processed includes described to be processed
Importance degree in comment information；
Word selecting module 403, for the sequence according to the importance degree being calculated from high to low, from described to be processed
In the word that comment information includes, selection has representational word to the comment information to be processed；
Sentiment orientation information acquisition module 404, for according to selected word, the emotion for obtaining the information to be analyzed to be inclined
To information.
In one embodiment of the application, the Sentiment orientation information acquisition module 404 is specifically used for:
Determine the emotion of selected word；
For identified each emotion, the quantity for belonging to the word of the emotion in selected word is counted；
According to the quantity that statistics obtains, the Sentiment orientation information of the information to be analyzed is obtained.
As seen from the above, when obtaining the Sentiment orientation information of information to be analyzed using scheme provided in this embodiment, due to
Above-mentioned comment information to be processed reflects Sentiment orientation of the reader to information to be analyzed of information to be analyzed, and by above-mentioned institute
The word of selection to comment information to be processed have it is representational, so, above-mentioned selected word is able to reflect readding for information to be analyzed
Sentiment orientation of the reader to information to be analyzed.Therefore, inclined according to the emotion that above-mentioned selected word can obtain information to be analyzed
To information.
In one embodiment of the application, referring to Fig. 5, the structure for providing second of Sentiment orientation information obtaining device is shown
It is intended to, compared with above-mentioned embodiment illustrated in fig. 4, in the present embodiment, above-mentioned importance degree computing module 402, comprising:
Word segmentation processing unit 402A obtains the first processing knot for carrying out word segmentation processing to the comment information to be processed
Fruit；
Stop words removal unit 402B obtains second processing knot for removing the stop words in first processing result
Fruit, wherein the stop words are as follows: preset word without emotion and/or do not have to content of text representational
Word；
Importance degree computing unit 402C to be processed is commented for calculating in the second processing result each word described
By the importance degree in information.
In one embodiment of the application, the importance degree computing unit 402C can be specifically used for:
The number that the sum and each word for counting word included in the second processing result occur；
For each word in the second processing result, the number of the information in preset information bank comprising the word is obtained
It measures, and calculates the first characterization value and the second characterization value of the word according to following formula:
Wherein, S1 indicates that first characterization value, S2 indicate that second characterization value, Num1 indicate the second processing knot
The number that the word occurs in fruit, Num2 indicate the sum of word included in the second processing result, in Num3 described information storehouse
The sum of included information, Num4 indicate the quantity of the information in described information storehouse comprising the word；
According to the first characterization value and the second characterization value of each word in the second processing result, calculates separately each word and exist
Importance degree in the comment information to be processed.
As seen from the above, when obtaining the Sentiment orientation information of information to be analyzed using scheme provided in this embodiment, due to
The stop words in comment information to be processed is eliminated, can to select has representational word to comment information to be processed
It is representational stronger, also more acurrate, so that the Sentiment orientation information of the information to be analyzed obtained is more accurate.
In one embodiment of the application, above-mentioned Sentiment orientation information obtaining device can also include:
Sample word obtains module, has in the word that the comment information for obtaining sample information is included to the sample information
There is representational sample word；
Coding result obtains module and obtains each sample word for encoding respectively to each sample word with binary system
The encoding samples result that numerical value indicates；
Vector obtain module, for according to the encoding samples as a result, generating the corresponding vector of each sample word, wherein
The corresponding vector of each sample word includes preset quantity binary numeral；
Model training module is instructed for the corresponding vector of each sample word to be inputted preset deep learning model
Practice, obtains Sentiment orientation and obtain model.
On the basis of obtaining above-mentioned Sentiment orientation acquisition model, referring to Fig. 6, the third Sentiment orientation information is provided
The structural schematic diagram for obtaining device, compared with above-mentioned embodiment illustrated in fig. 4, in the present embodiment, above-mentioned Sentiment orientation information acquisition
Module 404, comprising:
Coding result obtaining unit 404A is obtained selected each for encoding respectively to selected each word
The coding result that a word is indicated with binary numeral；
Vector generation unit 404B, for generating the corresponding vector of selected each word according to the coding result,
In, the corresponding vector of each selected word includes the preset quantity binary numeral；
Sentiment orientation information obtainment unit 404C, for the corresponding vector of selected each word to be input to the emotion
Tendency obtains model, obtains the Sentiment orientation information of the information to be analyzed.
As seen from the above, it when obtaining the Sentiment orientation information of information to be analyzed using scheme provided in this embodiment, is selecting
It selects after there is representational word to comment information to be processed, obtains model by means of Sentiment orientation trained in advance, obtain wait divide
Analyse the Sentiment orientation information of information.Since there are great amount of samples information, deep learning model can be very good to learn
The rule that the Sentiment orientation of sample information is shown is practised out, for this purpose, obtaining model by trained Sentiment orientation can be quasi-
The Sentiment orientation information of true acquisition information to be analyzed.
Corresponding with above-mentioned Sentiment orientation information obtaining method, the embodiment of the present application also provides a kind of electronic equipment.
Referring to Fig. 7, the structural schematic diagram of a kind of electronic equipment is provided, which includes: processor 701 and machine
Readable storage medium storing program for executing 702, the machine readable storage medium 702 is stored with the machine that can be executed by the processor 701 can
It executes instruction, the processor 701 is promoted by the machine-executable instruction: realizing Sentiment orientation described in the embodiment of the present application
Information obtaining method step.
In one embodiment of the application, a kind of Sentiment orientation information obtaining method is provided, this method comprises:
Comment information to be processed is obtained, the comment information to be processed is the comment information of information to be analyzed；
Calculate the importance degree of word that the comment information to be processed includes in the comment information to be processed；
According to the sequence of the importance degree being calculated from high to low, the word that includes from the comment information to be processed
In, selection has representational word to the comment information to be processed；
According to selected word, the Sentiment orientation information of the information to be analyzed is obtained.
It should be noted that above-mentioned processor 701 is promoted the Sentiment orientation information acquisition realized by machine-executable instruction
The other embodiments of method, identical as embodiment mentioned by preceding method embodiment part, which is not described herein again.
As seen from the above, when obtaining the Sentiment orientation information of information to be analyzed using electronic equipment provided in this embodiment,
Since above-mentioned comment information to be processed reflects Sentiment orientation of the reader to information to be analyzed of information to be analyzed, and due to upper
State selected word to comment information to be processed have it is representational, so, above-mentioned selected word is able to reflect information to be analyzed
Reader to the Sentiment orientation of information to be analyzed.Therefore, the feelings of information to be analyzed can be obtained according to above-mentioned selected word
Feel trend information.
Corresponding with above-mentioned Sentiment orientation information obtaining method, the embodiment of the present application also provides a kind of machine readable storages
Medium is stored with machine-executable instruction, and when being called and being executed by processor, the machine-executable instruction promotes the place
It manages device: realizing Sentiment orientation information obtaining method step described in the embodiment of the present application.
In one embodiment of the application, a kind of Sentiment orientation information obtaining method is provided, this method comprises:
Comment information to be processed is obtained, the comment information to be processed is the comment information of information to be analyzed；
Calculate the importance degree of word that the comment information to be processed includes in the comment information to be processed；
According to the sequence of the importance degree being calculated from high to low, the word that includes from the comment information to be processed
In, selection has representational word to the comment information to be processed；
According to selected word, the Sentiment orientation information of the information to be analyzed is obtained.
It should be noted that the Sentiment orientation information obtaining method that above-mentioned machine-executable instruction promotes processor to realize
Other embodiments, identical as embodiment mentioned by preceding method embodiment part, which is not described herein again.
As seen from the above, the machine-executable instruction stored in machine readable storage medium provided in this embodiment is executed to obtain
When obtaining the Sentiment orientation information of information to be analyzed, since above-mentioned comment information to be processed reflects the reader couple of information to be analyzed
The Sentiment orientation of information to be analyzed, but due to above-mentioned selected word to comment information to be processed have it is representational, so, it is above-mentioned
Selected word is able to reflect Sentiment orientation of the reader to information to be analyzed of information to be analyzed.Therefore, according to above-mentioned selected
The word selected can obtain the Sentiment orientation information of information to be analyzed.
It should be noted that above-mentioned machine readable storage medium may include random access memory (Random Access
Memory, RAM), it also may include nonvolatile memory (Non-Volatile Memory, NVM), for example, at least a magnetic
Disk storage.Optionally, above-mentioned machine readable storage medium can also be that at least one is located remotely from the storage of aforementioned processor
Device.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
Herein, relational terms such as first and second and the like be used merely to by an entity or operation with it is another
One entity or operation distinguish, and without necessarily requiring or implying between these entities or operation, there are any this reality
Relationship or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to the packet of nonexcludability
Contain, so that the process, method, article or equipment for including a series of elements not only includes those elements, but also including
Other elements that are not explicitly listed, or further include for elements inherent to such a process, method, article, or device.
In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including the element
Process, method, article or equipment in there is also other identical elements.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device,
For electronic equipment and machine readable storage medium embodiment, since it is substantially similar to the method embodiment, so the ratio of description
Relatively simple, the relevent part can refer to the partial explaination of embodiments of method.
The foregoing is merely the preferred embodiments of the application, are not intended to limit the protection scope of the application.It is all
Any modification, equivalent replacement, improvement and so within spirit herein and principle are all contained in the protection scope of the application
It is interior.
Claims (12)
1. a kind of Sentiment orientation information obtaining method, which is characterized in that the described method includes:
Comment information to be processed is obtained, the comment information to be processed is the comment information of information to be analyzed；
Calculate the importance degree of word that the comment information to be processed includes in the comment information to be processed；
According to the sequence of the importance degree being calculated from high to low, from the word that the comment information to be processed includes, choosing
Selecting has representational word to the comment information to be processed；
According to selected word, the Sentiment orientation information of the information to be analyzed is obtained.
2. the method according to claim 1, wherein the word that the calculating comment information to be processed includes exists
Importance degree in the comment information to be processed, comprising:
Word segmentation processing is carried out to the comment information to be processed, obtains the first processing result；
The stop words in first processing result is removed, obtains second processing result, wherein the stop words are as follows: set in advance
Fixed word without emotion and/or do not have representational word to content of text；
Calculate importance degree of each word in the comment information to be processed in the second processing result.
3. according to the method described in claim 2, it is characterized in that, each word is in institute in the calculating second processing result
State the importance degree in comment information to be processed, comprising:
The number that the sum and each word for counting word included in the second processing result occur；
For each word in the second processing result, the quantity of the information in preset information bank comprising the word is obtained, and
The first characterization value and the second characterization value of the word are calculated according to following formula:
Wherein, S1 indicates that first characterization value, S2 indicate that second characterization value, Num1 indicate in the second processing result
The number that the word occurs, Num2 indicate the sum of word included in the second processing result, are wrapped in Num3 described information storehouse
Sum containing information, Num4 indicate the quantity of the information in described information storehouse comprising the word；
According to the first characterization value and the second characterization value of each word in the second processing result, each word is calculated separately described
Importance degree in comment information to be processed.
4. method according to any one of claim 1-3, which is characterized in that the method also includes:
There is representational sample word to the sample information in the word that the comment information of acquisition sample information is included；
Each sample word is encoded respectively, obtains the encoding samples result that each sample word is indicated with binary numeral；
According to the encoding samples as a result, generating the corresponding vector of each sample word, wherein the corresponding vector packet of each sample word
Containing preset quantity binary numeral；
The corresponding vector of each sample word is inputted preset deep learning model to be trained, Sentiment orientation is obtained and obtains mould
Type；
It is described according to selected word, obtain the Sentiment orientation information of the information to be analyzed, comprising:
Selected each word is encoded respectively, obtains the coding knot that selected each word is indicated with binary numeral
Fruit；
According to the coding result, generate the corresponding vector of selected each word, wherein each selected word it is corresponding to
Amount includes the preset quantity binary numeral；
The corresponding vector of selected each word is input to the Sentiment orientation and obtains model, obtains the information to be analyzed
Sentiment orientation information.
5. method according to any one of claim 1-3, which is characterized in that it is described according to selected word, obtain institute
State the Sentiment orientation information of information to be analyzed, comprising:
Determine the emotion of selected word；
For identified each emotion, the quantity for belonging to the word of the emotion in selected word is counted；
According to the quantity that statistics obtains, the Sentiment orientation information of the information to be analyzed is obtained.
6. a kind of Sentiment orientation information obtaining device, which is characterized in that described device includes:
Comment information obtains module, and for obtaining comment information to be processed, the comment information to be processed is information to be analyzed
Comment information；
Importance degree computing module, for calculating word that the comment information to be processed includes in the comment information to be processed
In importance degree；
Word selecting module is believed for the sequence according to the importance degree being calculated from high to low from the comment to be processed
In the word that breath includes, selection has representational word to the comment information to be processed；
Sentiment orientation information acquisition module, for obtaining the Sentiment orientation information of the information to be analyzed according to selected word.
7. device according to claim 6, which is characterized in that the importance degree computing module, comprising:
Word segmentation processing unit obtains the first processing result for carrying out word segmentation processing to the comment information to be processed；
Stop words removal unit obtains second processing result for removing the stop words in first processing result, wherein
The stop words are as follows: preset word without emotion and/or do not have representational word to content of text；
Importance degree computing unit, for calculating in the second processing result each word in the comment information to be processed
Importance degree.
8. device according to claim 7, which is characterized in that the importance degree computing unit is specifically used for:
The number that the sum and each word for counting word included in the second processing result occur；
For each word in the second processing result, the quantity of the information in preset information bank comprising the word is obtained, and
The first characterization value and the second characterization value of the word are calculated according to following formula:
Wherein, S1 indicates that first characterization value, S2 indicate that second characterization value, Num1 indicate in the second processing result
The number that the word occurs, Num2 indicate the sum of word included in the second processing result, are wrapped in Num3 described information storehouse
Sum containing information, Num4 indicate the quantity of the information in described information storehouse comprising the word；
According to the first characterization value and the second characterization value of each word in the second processing result, each word is calculated separately described
Importance degree in comment information to be processed.
9. device a method according to any one of claims 6-8, which is characterized in that described device further include:
Sample word obtains module, has table to the sample information in the word that the comment information for obtaining sample information is included
The sample word of sign property；
Coding result obtains module and obtains each sample word for encoding respectively to each sample word with binary numeral
The encoding samples result of expression；
Vector obtains module, for according to the encoding samples as a result, generating the corresponding vector of each sample word, wherein it is each
The corresponding vector of sample word includes preset quantity binary numeral；
Model training module is trained for the corresponding vector of each sample word to be inputted preset deep learning model, obtains
It obtains Sentiment orientation and obtains model；
The Sentiment orientation information acquisition module, comprising:
Coding result obtaining unit obtains selected each word for encoding respectively to selected each word with two
The coding result that binary value indicates；
Vector generation unit, for generating the corresponding vector of selected each word, wherein each according to the coding result
The corresponding vector of selected word includes the preset quantity binary numeral；
Sentiment orientation information obtainment unit is obtained for the corresponding vector of selected each word to be input to the Sentiment orientation
Model obtains the Sentiment orientation information of the information to be analyzed.
10. device a method according to any one of claims 6-8, which is characterized in that the Sentiment orientation information acquisition module,
It is specifically used for:
Determine the emotion of selected word；
For identified each emotion, the quantity for belonging to the word of the emotion in selected word is counted；
According to the quantity that statistics obtains, the Sentiment orientation information of the information to be analyzed is obtained.
11. a kind of electronic equipment, which is characterized in that including processor and machine readable storage medium, the machine readable storage
Media storage has the machine-executable instruction that can be executed by the processor, and the processor is by the machine-executable instruction
Promote: realizing any method and step of claim 1-5.
12. a kind of machine readable storage medium, which is characterized in that be stored with machine-executable instruction, by processor call and
When execution, the machine-executable instruction promotes the processor: realizing any method and step of claim 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910807068.3A CN110516249A (en) | 2019-08-29 | 2019-08-29 | A kind of Sentiment orientation information obtaining method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910807068.3A CN110516249A (en) | 2019-08-29 | 2019-08-29 | A kind of Sentiment orientation information obtaining method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110516249A true CN110516249A (en) | 2019-11-29 |
Family
ID=68628844
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910807068.3A Pending CN110516249A (en) | 2019-08-29 | 2019-08-29 | A kind of Sentiment orientation information obtaining method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110516249A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111565322A (en) * | 2020-05-14 | 2020-08-21 | 北京奇艺世纪科技有限公司 | User emotional tendency information obtaining method and device and electronic equipment |
| WO2022116155A1 (en) * | 2020-12-04 | 2022-06-09 | 中国科学院深圳先进技术研究院 | Method for determining emotion processing tendency, and related product |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017149540A1 (en) * | 2016-03-02 | 2017-09-08 | Feelter Sales Tools Ltd | Sentiment rating system and method |
| CN108197104A (en) * | 2017-12-27 | 2018-06-22 | 浙江力石科技股份有限公司 | Text analyzing method, apparatus and cloud platform |
| CN108197109A (en) * | 2017-12-29 | 2018-06-22 | 北京百分点信息科技有限公司 | A kind of multilingual analysis method and device based on natural language processing |
| CN108460010A (en) * | 2018-01-17 | 2018-08-28 | 南京邮电大学 | A kind of comprehensive grade model implementation method based on sentiment analysis |
| CN109977206A (en) * | 2019-03-20 | 2019-07-05 | 西北大学 | A kind of short text feature extracting method blended based on multiple features factor |
-
2019
- 2019-08-29 CN CN201910807068.3A patent/CN110516249A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017149540A1 (en) * | 2016-03-02 | 2017-09-08 | Feelter Sales Tools Ltd | Sentiment rating system and method |
| CN108197104A (en) * | 2017-12-27 | 2018-06-22 | 浙江力石科技股份有限公司 | Text analyzing method, apparatus and cloud platform |
| CN108197109A (en) * | 2017-12-29 | 2018-06-22 | 北京百分点信息科技有限公司 | A kind of multilingual analysis method and device based on natural language processing |
| CN108460010A (en) * | 2018-01-17 | 2018-08-28 | 南京邮电大学 | A kind of comprehensive grade model implementation method based on sentiment analysis |
| CN109977206A (en) * | 2019-03-20 | 2019-07-05 | 西北大学 | A kind of short text feature extracting method blended based on multiple features factor |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111565322A (en) * | 2020-05-14 | 2020-08-21 | 北京奇艺世纪科技有限公司 | User emotional tendency information obtaining method and device and electronic equipment |
| CN111565322B (en) * | 2020-05-14 | 2022-03-04 | 北京奇艺世纪科技有限公司 | User emotional tendency information obtaining method and device and electronic equipment |
| WO2022116155A1 (en) * | 2020-12-04 | 2022-06-09 | 中国科学院深圳先进技术研究院 | Method for determining emotion processing tendency, and related product |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108363790B (en) | Method, device, equipment and storage medium for evaluating comments | |
| KR102431549B1 (en) | Causality recognition device and computer program therefor | |
| CN108319668A (en) | Generate the method and apparatus of text snippet | |
| CN110032633B (en) | Multi-turn dialogue processing method, device and equipment | |
| CN111368042A (en) | Intelligent question and answer method and device, computer equipment and computer storage medium | |
| CN111144127B (en) | Text semantic recognition method, text semantic recognition model acquisition method and related device | |
| CN109710947A (en) | Power specialty word stock generating method and device | |
| CN109325109A (en) | Attention encoder-based extraction type news abstract generating device | |
| CN112528637A (en) | Text processing model training method and device, computer equipment and storage medium | |
| CN109034203A (en) | Training, expression recommended method, device, equipment and the medium of expression recommended models | |
| CN105279227A (en) | Voice search processing method and device of homonym | |
| CN110210038B (en) | Core entity determining method, system, server and computer readable medium thereof | |
| CN110516249A (en) | A kind of Sentiment orientation information obtaining method and device | |
| Wohlgenannt et al. | Extracting social networks from literary text with word embedding tools | |
| CN111125384B (en) | Multimedia answer generation method and device, terminal equipment and storage medium | |
| JP5973935B2 (en) | Browsing behavior prediction device, browsing behavior prediction method, and program | |
| CN108595395B (en) | Nickname generation method, device and equipment | |
| CN110413750A (en) | The method and apparatus for recalling standard question sentence according to user's question sentence | |
| CN112100986B (en) | Voice text clustering method and device | |
| CN117556263B (en) | Sample construction method, code generation method, electronic device, and storage medium | |
| CN111898389B (en) | Information determination method, information determination device, computer equipment and storage medium | |
| EP4113289A1 (en) | Automated code analysis and tagging (methods and systems) | |
| CN109783800B (en) | Emotion keyword acquisition method, device, equipment and storage medium | |
| CN115577084A (en) | Conversation strategy prediction method and prediction device | |
| Lasy et al. | Dialogue System Augmented with Commonsense Knowledge |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191129 |
|
| RJ01 | Rejection of invention patent application after publication |