CN112651237B - User portrait establishing method and device based on user emotion standpoint and user portrait visualization method - Google Patents
User portrait establishing method and device based on user emotion standpoint and user portrait visualization method Download PDFInfo
- Publication number
- CN112651237B CN112651237B CN201910961379.5A CN201910961379A CN112651237B CN 112651237 B CN112651237 B CN 112651237B CN 201910961379 A CN201910961379 A CN 201910961379A CN 112651237 B CN112651237 B CN 112651237B
- Authority
- CN
- China
- Prior art keywords
- emotion
- words
- word
- user
- probability
- 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.)
- Active
Links
- 230000008451 emotion Effects 0.000 title claims abstract description 548
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000007794 visualization technique Methods 0.000 title description 7
- 238000012937 correction Methods 0.000 claims description 30
- 230000006397 emotional response Effects 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 17
- 238000012216 screening Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 10
- 230000007935 neutral effect Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 7
- 238000012935 Averaging Methods 0.000 claims description 5
- 239000000463 material Substances 0.000 claims description 5
- 230000001960 triggered effect Effects 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 abstract description 14
- 238000010586 diagram Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 9
- 238000010606 normalization Methods 0.000 description 8
- 239000003550 marker Substances 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 238000012986 modification Methods 0.000 description 6
- 230000004048 modification Effects 0.000 description 6
- 238000012800 visualization Methods 0.000 description 6
- 235000015895 biscuits Nutrition 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 208000027534 Emotional disease Diseases 0.000 description 4
- 241000272470 Circus Species 0.000 description 3
- 230000006399 behavior Effects 0.000 description 3
- 238000000546 chi-square test Methods 0.000 description 3
- 239000003086 colorant Substances 0.000 description 3
- 241001465754 Metazoa Species 0.000 description 2
- 241000237536 Mytilus edulis Species 0.000 description 2
- 239000008186 active pharmaceutical agent Substances 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 235000020638 mussel Nutrition 0.000 description 2
- 230000004083 survival effect Effects 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 108700019146 Transgenes Proteins 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000002996 emotional effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
Abstract
The invention discloses a user portrait establishing method and device based on a user emotion standpoint, and a user portrait visualizing method, wherein the user portrait establishing method comprises the following steps: acquiring independent short text corpus from user history data; classifying the obtained short text corpus according to emotion tendencies, and constructing an emotion word library according to the distribution condition of words in the short text corpus classification result; constructing a standing trigger word library according to an application scene; calculating emotion probability of the speech block to be analyzed; according to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the standing trigger word corresponding to the single user; calculating the average emotion probability of the standing trigger words in the communities according to the emotion probabilities of the standing trigger words corresponding to the single users, and sorting according to the average emotion probability; and constructing user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words. The method can improve the accuracy and intuitiveness of emotion analysis of the user.
Description
Technical Field
The invention relates to the technical field of data analysis, in particular to a user portrait establishing method and device based on a user emotion standpoint and a user portrait visualization method.
Background
User behavior of a user in a network platform is often used to describe a user's characteristics, which are referred to as user portraits, and the emphasis on constructing user portraits may vary depending on the purpose. For example, the e-commerce platform focuses on the consumption capability of the user, the purchasing preference establishes a user portrait, the social platform establishes the user portrait based on the interest characteristics and social relations of the user, and different user portraits help the platform to classify the user, so that customized services are better realized for the user.
In the process of implementing the present invention, the present inventors have found that the method of the prior art has at least the following technical problems:
in the prior art, when emotion analysis is performed on a user, a large number of non-network words and non-daily words are contained in an emotion word library adopted by the user, and words of common network words at present are lacking, so that the emotion analysis accuracy and the practicability based on the existing emotion word library are limited.
From this, it is known that the method in the prior art has a technical problem that the analysis result is not accurate enough.
Disclosure of Invention
In view of the above, the invention provides a user portrait establishing method and device based on a user emotion standpoint, and a user portrait visualizing method, which are used for solving or at least partially solving the technical problem that the result of the method in the prior art is not accurate enough.
The first aspect of the invention provides a user portrait establishment method based on a user emotion standpoint, comprising the following steps:
acquiring independent short text corpus from user history data;
classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word library according to the distribution condition of words in the short text corpus classification result, and calculating the original emotion probability of emotion words in the emotion word library; wherein, the emotion word library comprises positive emotion words and negative emotion words;
constructing a standing triggering word library according to an application scene, wherein the standing triggering word library comprises standing triggering words which can cause standing or emotional response of a user;
extracting standing trigger words contained in text information issued by a user, forming a to-be-analyzed text block according to the extracted standing trigger words, and calculating the emotion probability of the to-be-analyzed text block according to the emotion probability of emotion words in the to-be-analyzed text block, the number of degree adverbs and the number of negative words;
According to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the standing trigger word corresponding to the single user;
calculating the average emotion probability of the standing trigger words in the community formed by all the users according to the emotion probability of the standing trigger words corresponding to the single user, and sorting according to the average emotion probability;
and constructing user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words.
In one embodiment, classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word bank according to the distribution condition of words in the short text corpus classification result, and calculating emotion probability of emotion words in the emotion word bank, including:
classifying the obtained short text corpus according to emotion tendencies, and dividing the short text corpus into positive corpus, neutral corpus and negative corpus;
dividing the classified corpus into words and removing redundancy to obtain a corpus word library;
counting the distribution condition of each word in the corpus word library in positive corpus, neutral corpus and negative corpus;
according to the distribution condition of words, screening words related to positive and negative directions by combining chi-square verification as mark word candidates of emotion tendencies;
Screening the mark word candidates, deleting words which are not matched with the corresponding emotion tendencies, and constructing an emotion word library;
searching all original corpus corresponding to each positive emotion word, calculating the average value of the positive emotion probabilities, taking the average value as the original emotion probability of the positive emotion word, and subtracting the average value of the positive emotion probabilities from 1 for negative emotion words in an emotion word bank to obtain the original emotion probability of the negative emotion word.
In one embodiment, forming a to-be-analyzed speech block according to the extracted standing trigger words, and then calculating the emotion probability of the to-be-analyzed speech block according to the number of emotion words, degree adverbs and negatives in the to-be-analyzed speech block, wherein the method comprises the following steps:
forming sentences in which the extracted standing trigger words are located and n sentences before and after the sentences to form a to-be-analyzed language block, wherein n is a positive integer greater than or equal to 1;
searching positive emotion words and negative emotion words appearing in the to-be-analyzed language material block, and acquiring original emotion probability of each positive emotion word and negative emotion word;
determining a negative coefficient and a degree weight according to the number of negative words and degree adverbs in a preset range and according to each positive emotion word and each negative emotion word;
Calculating an emotion probability correction value of each emotion word according to the original emotion probability, the negative coefficient and the degree weight of the emotion word;
and calculating the emotion probability of the to-be-analyzed language block according to the emotion probability correction value of the emotion words, the number of positive emotion words and the number of negative emotion words.
In one embodiment, according to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the standing trigger word corresponding to the single user includes:
when the standing trigger word does not appear in the data issued by the user, the emotion probability of the standing trigger word is null;
when the standing trigger word appears once in the data issued by the user, the emotion probability corresponding to the corpus where the standing trigger word appears is used as the emotion probability of the standing trigger word;
when the standing trigger word appears twice or more in the data issued by the user, taking the average value of the emotion probabilities corresponding to all corpus blocks where the standing trigger word appears as the emotion probability of the standing trigger word.
In one embodiment, after calculating the emotion probability of the position trigger word corresponding to the single user according to the emotion probability of the to-be-analyzed language block, the method further includes:
and normalizing the emotion probability of the standing trigger words corresponding to the single user to obtain the emotion probability correction value of each standing trigger word corresponding to the single user.
In one embodiment, calculating the average emotion probability of the standpoint trigger words in the community composed of all users according to the emotion probability of the standpoint trigger words corresponding to the single user comprises:
averaging emotion probability correction values of standing trigger words triggered by standing corresponding to each user, and calculating average emotion probability of the standing trigger words in the community;
and sorting according to the average emotion probability of each standing trigger word.
Based on the same inventive concept, a second aspect of the present invention provides a user portrait creation apparatus based on a user emotion standpoint, including:
the corpus acquisition module is used for acquiring independent short text corpus from the user history data;
the emotion word bank construction module is used for classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word bank according to the distribution condition of words in the short text corpus classification result and calculating the original emotion probability of emotion words in the emotion word bank; wherein, the emotion word library comprises positive emotion words and negative emotion words;
the establishment triggering word stock construction module is used for constructing an establishment triggering word stock according to an application scene, wherein the establishment triggering word stock comprises establishment triggering words capable of causing establishment or emotional response of a user;
The corpus block emotion probability calculation module is used for extracting standing trigger words contained in text information issued by a user, forming a corpus block to be analyzed according to the extracted standing trigger words, and calculating the emotion probability of the corpus block to be analyzed according to the emotion probability of the emotion words in the corpus block to be analyzed, the number of degree auxiliary words and the number of negative words;
the single user standing trigger word emotion probability calculation module is used for calculating the emotion probability of the standing trigger word corresponding to the single user according to the emotion probability of the to-be-analyzed language block;
the average emotion probability ordering module is used for calculating the average emotion probability of the standing trigger words in the community formed by all the users according to the emotion probability of the standing trigger words corresponding to the single user, and ordering according to the average emotion probability;
and the user portrait construction module constructs user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words.
Based on the same inventive concept, a third aspect of the present invention provides a user portrait visualization method, including: and visually displaying the user portrait constructed by the method of the first aspect.
In one embodiment, visually displaying a representation of a user includes:
mapping the standing trigger words to word blocks with preset shapes according to the average emotion probability;
constructing a corresponding relation between the emotion probability of a single user to the standing trigger word and the color characteristics;
and visually displaying the user portrait according to the corresponding relation between the emotion probability and the color characteristics.
Based on the same inventive concept, a fourth aspect of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed, implements the method of the first aspect
The above-mentioned one or more technical solutions in the embodiments of the present application at least have one or more of the following technical effects:
the invention provides a user portrait establishing method based on a user emotion standpoint, which comprises the steps of firstly, obtaining independent short text corpus from user history data; then constructing an emotion word library, and calculating the original emotion probability of emotion words in the emotion word library; then constructing a standing trigger word library according to the application scene; then calculating the emotion probability of the speech block to be analyzed; according to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the standing trigger word corresponding to the single user; then calculating the average emotion probability of the standing trigger words in the community formed by all the users, and sorting according to the average emotion probability; and finally, constructing user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words.
According to the method provided by the invention, the emotion word library can be constructed according to the distribution condition of words in the short text corpus classification result, the constructed emotion word library is divided into positive emotion words and negative emotion words according to emotion tendencies, emotion tendencies of new words can be better identified from text contents of users, emotion reactions in the new words can be accurately analyzed, a standing trigger word library is constructed, mapping between a single user and the standing trigger word library is then established, and emotion probability of standing trigger words corresponding to the single user is calculated; and then calculating the average emotion probability of the standing trigger words in the community formed by all the users, and sorting the average emotion probability, so that user portraits can be constructed according to the sorting result, the emotion standing of each user on the standing trigger words can be accurately realized, the viewpoint difference and emotion response characteristics of each user on common things can be rapidly and accurately known, and the technical problem that the analysis result is inaccurate in the method in the prior art is solved.
Furthermore, based on the constructed user portrait, the invention also provides a visualization method of the user portrait, which is used for carrying out visualization display on the user portrait and improving intuitiveness.
Further, according to the color gradient formula, the corresponding relation between the emotion probability of the single user to the standing trigger words and the color characteristics is constructed, and then the user portrait is visually displayed according to the corresponding relation between the emotion probability and the color characteristics, so that the display effect can be improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a user portrayal creation method based on a user emotion standpoint of the present invention;
FIG. 2 is a block diagram of a user portrayal creation device based on a user emotion standpoint in an embodiment of the present invention;
FIG. 3 is a diagram showing a user portrait visualization effect of an individual user in an embodiment of the present invention;
FIG. 4 is a diagram showing user portrayal visualization effects for different users in an embodiment of the present invention;
FIG. 5 is a diagram showing a user portrait visualization effect of a group of users in an embodiment of the present invention;
fig. 6 is a block diagram of a computer-readable storage medium according to an embodiment of the present invention.
Detailed Description
Aiming at the technical problem that the result of the method in the prior art is not accurate enough, the invention provides a user portrait establishing method and device based on user emotion, and a user portrait visualizing method.
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Through a great deal of research and practice, the inventor of the application finds that in the behavior of the user, a lot of data can reflect the emotion of the user, the emotion corresponds to the viewpoint and the standpoint of the user on a lot of things, and the emotion map of the user can be constructed based on the emotion response of the user on common things. In the aspect of emotion analysis, the existing emotion word library contains a large number of words of non-network words and non-daily words, and meanwhile, the words of common words in the current network are lacking, so that the accuracy and the practicability of emotion analysis based on the existing emotion word library are limited.
The invention provides a user portrait establishing method and a user portrait visualizing method based on a user emotion standpoint, wherein a novel emotion word library establishing method is adopted, and the overall method has the following advantages or beneficial technical effects:
1. by adopting the emotion word library constructed in the invention, the emotion tendency of the new word can be better identified from the text content of the user, and the emotion response in the emotion word library can be accurately analyzed;
2. so that the user can quickly know the difference of the perspectives and the emotional response characteristics of the user and other people to common things;
3. visual analysis is carried out on the emotion standing characteristics of a certain user group;
4. And at the same time, the classification of the user in more dimensions is facilitated.
Example 1
The embodiment provides a user portrait establishing method based on a user emotion standpoint, referring to fig. 1, the method includes:
step S1: and obtaining independent short text corpus from the user history data.
Specifically, the user history data includes history behavior data of the user, such as message information of the user, posted comment information, and the like. The independent short text corpus represents texts with certain meanings, and the text can be realized through the existing tools.
Step S2: classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word library according to the distribution condition of words in the short text corpus classification result, and calculating the original emotion probability of emotion words in the emotion word library; wherein the emotion word library comprises positive emotion words and negative emotion words.
Specifically, emotional tendency is the position or attitude of the user, e.g., "like" means positive or positive emotion, "eat" means neutral, "offence" means negative or negative emotion. The words, namely the words contained in the classified corpus, can be obtained through word segmentation operation. The positive emotion probability of the original corpus is the probability of positive emotion of emotion contained in the corpus calculated by adopting an open source emotion analysis API. Searching all original corpus corresponding to each emotion word, and calculating the average value of the positive emotion probabilities of the original corpus and the emotion word, so that the original emotion probability of the emotion word can be obtained.
Step S3: and constructing a standing triggering word library according to the application scene, wherein the standing triggering word library comprises standing triggering words which can cause standing or emotional response of the user.
In particular, nouns, such as songs, movies, celebrities, popular concepts, etc., gathered from application scenes and platform features that may elicit a user's position or emotional response. The user can more specifically indicate whether the attitude to a song, a movie, a persona or a popular concept such as "transgene" is positive or negative. And the set of standpoint trigger words is a library of standpoint trigger words.
Step S4: extracting standing trigger words contained in text information issued by a user, forming a to-be-analyzed text block according to the extracted standing trigger words, and calculating the emotion probability of the to-be-analyzed text block according to the emotion probability of emotion words in the to-be-analyzed text block, the number of degree adverbs and the number of negative words.
Specifically, the text information posted by the user may be comments, articles, and other contents. In the implementation, sentences where the standpoint trigger words are located and context information can be combined to form the to-be-analyzed language block.
Step S5: and calculating the emotion probability of the standing trigger word corresponding to the single user according to the emotion probability of the to-be-analyzed language block.
Specifically, the emotion probability of the standing trigger word corresponding to the single user can be determined according to the occurrence times, the occurrence time and the like of the standing trigger word in the corpus to be analyzed, so that the mapping between the single user and the standing trigger word is constructed.
Step S6: according to the emotion probabilities of the standing trigger words corresponding to the single user, calculating the average emotion probabilities of the standing trigger words in the community formed by all the users, and sorting according to the average emotion probabilities.
Specifically, the emotion probabilities of the standing trigger words in the standing trigger word library are obtained in step S5 for the individual users, and the average emotion probability of the standing trigger words in the community formed by all the users is obtained in this step. The emotion probability sum of all the corpus in which the standing trigger word appears can be calculated by dividing the emotion probability sum by the total corpus text number in which the word appears.
Step S7: and constructing user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words.
Specifically, the user portraits can be constructed according to the average emotion probabilities of the standing trigger words from large to small and then according to the emotion probabilities of the individual users to the standing trigger words.
In one embodiment, classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word bank according to the distribution condition of words in the short text corpus classification result, and calculating emotion probability of emotion words in the emotion word bank, including:
classifying the obtained short text corpus according to emotion tendencies, and dividing the short text corpus into positive corpus, neutral corpus and negative corpus;
dividing the classified corpus into words and removing redundancy to obtain a corpus word library;
counting the distribution condition of each word in the corpus word library in positive corpus, neutral corpus and negative corpus;
according to the distribution condition of words, screening words related to positive and negative directions by combining chi-square verification as mark word candidates of emotion tendencies;
screening the mark word candidates, deleting words which are not matched with the corresponding emotion tendencies, and constructing an emotion word library;
searching all original corpus corresponding to each positive emotion word, calculating the average value of the positive emotion probabilities, taking the average value as the original emotion probability of the positive emotion word, and subtracting the average value of the positive emotion probabilities from 1 for negative emotion words in an emotion word bank to obtain the original emotion probability of the negative emotion word.
Specifically, when screening candidate marker words, the chi-square test is used to judge the marker effect of each word on three types of corpus, namely whether the word is contained in sentences or not and whether the three emotion tendencies of positive, neutral and negative are related or not, and then the chi-square test result is used to screen candidate marker words. The marker word candidates refer to a marker word set which is preliminarily screened out.
And performing part-of-speech screening and manual screening on the candidate marker words, removing words which are not matched with the corresponding emotion tendencies, taking the rest words as emotion words, and dividing the emotion tendencies corresponding to each word into positive emotion words and negative emotion words.
And then, respectively calculating the original emotion probabilities of the positive emotion words and the negative emotion words, searching all original corpus corresponding to each word in the word stock for the positive emotion word stock, and calculating the average value of the positive emotion probabilities of the words as the original emotion probability of the word. In order to facilitate calculation, a normalization formula is utilized, and the value range of the original emotion probability of all words in the front emotion word library is unified to be between 0 and 1 and is used as emotion scores of the words. (Note that the positive emotion probability of the original corpus refers to the probability of positive emotion of emotion contained in the corpus calculated by using an open source emotion analysis API)
For the negative emotion word bank, searching all original linguistic data corresponding to each word in the word bank, subtracting the average value of the positive emotion probabilities of all the original linguistic data by 1, and taking the result as the original emotion probability of the word. For the convenience of calculation, the value ranges of the emotion probabilities of all words in the negative emotion word library are unified to be between 0 and 1 by using a normalization formula, and the value ranges are used as emotion scores of the words.
Wherein, the normalized formula is: x' = (X-X) min )/(X max -X min ) Wherein X represents data to be normalized, X' is a value obtained by normalizing X, and X max Is the maximum value of all data needing normalization, X min Is the minimum of all the data that needs to be normalized.
Through the method, the emotion word stock with emotion tendency can be constructed, and candidate marker words can be screened out more accurately by adopting the chi-square test method, so that the emotion word stock is more accurate.
In one embodiment, forming a to-be-analyzed speech block according to the extracted standing trigger words, and then calculating the emotion probability of the to-be-analyzed speech block according to the number of emotion words, degree adverbs and negatives in the to-be-analyzed speech block, wherein the method comprises the following steps:
Forming sentences in which the extracted standing trigger words are located and n sentences before and after the sentences to form a to-be-analyzed language block, wherein n is a positive integer greater than or equal to 1;
searching positive emotion words and negative emotion words appearing in the to-be-analyzed language material block, and acquiring original emotion probability of each positive emotion word and negative emotion word;
determining a negative coefficient and a degree weight according to the number of negative words and degree adverbs in a preset range and according to each positive emotion word and each negative emotion word;
calculating an emotion probability correction value of each emotion word according to the original emotion probability, the negative coefficient and the degree weight of the emotion word;
and calculating the emotion probability of the to-be-analyzed language block according to the emotion probability correction value of the emotion words, the number of positive emotion words and the number of negative emotion words.
Specifically, the to-be-analyzed language block is composed of 2n+1 sentences in total of the sentence where the standing trigger word is located and the n sentences before and after. And then analyzing the emotion characteristics of the speech block to be analyzed.
In a specific implementation process, the preset range can be selected according to needs, for example, a plurality of words appearing before and after the emotion word can be selected. For example, the implementation is as follows:
a. from each positive emotion word and negative emotion word, look forward for two words, if the two words contain k negative words (k has a value of [0,2 ] ]) Negative coefficient N of the emotion vocabulary i ＝(-1) k 。
b. If the two words do not contain the degreeAdverbs, the degree coefficient L of the emotion vocabulary i ＝1；
c. If 1 degree adverb is contained, the degree weight of the degree adverb is L, and the degree coefficient of the emotion word is L i If the expression level contains 2 degree adverbs, the degree weights are respectively L 1 ,L 2 The degree coefficient L of the emotion word i ＝L 1 ×L 2 ；
d. Calculating emotion probability correction value P of each emotion word i_index ＝P i ×N i ×L i 。
And then calculating the emotion probability of the to-be-analyzed voice block according to the calculated emotion probability correction value of the emotion words, the number of positive emotion words and the number of negative emotion words.
In specific implementation, the emotion word score emotion probability correction value is divided into two groups of positive emotion words and negative emotion words according to emotion types, namely, the emotion probability correction value is larger than or equal to 1 and is used as the positive emotion word, the emotion probability correction value is smaller than 1 and is used as the negative emotion word, and the emotion word score emotion probability correction value is arranged from large to small according to the numerical value. Wherein the number of positive emotion words is N p The number of negative emotion words is N n . The emotion score S for the chunk is then calculated as follows:
if N p ≥N n And N n Not equal to 0, firstly calculate the average value of all negative emotion wordsAnd calculate the previous N n Mean value of the positive emotion words ∈ - >And the average value of the remaining positive emotion words +.>Then calculate +.>And->Mean value of>Emotion score of a chunk->
If N p <N n And N p Not equal to 0, firstly calculate the average value of all the front emotion wordsN before calculation n Average value of individual negative emotion words +.>And the average value of the remaining negative emotion words +.>Then calculate +.>And->Mean value of>Emotion score of a chunk->
If N n And if the number of the negative emotion words in the corpus is 0, the negative emotion words in the corpus are all positive emotion words. Calculating the average value of all positive emotion wordsEmotion score of a chunk->Wherein L is max Is the maximum value of the range weights in the range adverb lexicon.
If N p And if the total emotion vocabulary is equal to 0, the number of positive emotion vocabularies in the corpus is 0, and all negative emotion vocabularies in the corpus are negative emotion vocabularies. Calculating the average of all negative emotion wordsEmotion score of a chunk->Wherein L is max Is the maximum value of the degree weight in the degree adverb word stock.
The objectivity and accuracy of calculation can be improved by combining the emotion probability correction value of the emotion words, the number of positive emotion words and the number of negative emotion words.
In one embodiment, according to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the standing trigger word corresponding to the single user includes:
When the standing trigger word does not appear in the data issued by the user, the emotion probability of the standing trigger word is null;
when the standing trigger word appears once in the data issued by the user, the emotion probability corresponding to the corpus where the standing trigger word appears is used as the emotion probability of the standing trigger word;
when the standing trigger word appears twice or more in the data issued by the user, taking the average value of the emotion probabilities corresponding to all corpus blocks where the standing trigger word appears as the emotion probability of the standing trigger word.
Specifically, for the standing trigger words appearing twice or more in the user data, the emotion scores thereof are calculated according to a weighted average of the chronological order of the occurrence of the corpus blocks containing the emotion trigger words.
For example, assuming that a user appears n blocks of speech for a standing trigger word together, the time span is T, dividing T by 10 to obtain 10 intervals, T1-T10, wherein the weight of each interval is 1-10, n blocks of speech are distributed in 10 intervals, the emotion probability of each block of speech is multiplied by the weight of the corresponding interval, and finally weighted average is taken to obtain the emotion probability of the corresponding standing trigger word.
In one embodiment, after calculating the emotion probability of the position trigger word corresponding to the single user according to the emotion probability of the to-be-analyzed language block, the method further includes:
And normalizing the emotion probability of the standing trigger words corresponding to the single user to obtain the emotion probability correction value of each standing trigger word corresponding to the single user.
Specifically, in order to facilitate subsequent calculation, the embodiment also normalizes the emotion probabilities of the standpoint trigger words corresponding to the individual users.
In a specific implementation process, for each user j of the user population to be analyzed, each standpoint trigger word k corresponding to the user j:
and respectively carrying out normalization processing on the positive emotion probabilities of all standing trigger words with non-null values (according to the calculated emotion probabilities of the standing trigger words and the emotion tendencies), so that the correction values of all the positive emotion probabilities are distributed between 0 and 1. For each emotion probability score S jk Its correction value S jk_index The calculation method of (1) is as follows, wherein S min Is the minimum value in all positive emotion probability scores, S max The maximum value of all positive emotion probability scores is represented by the following normalization processing formula: s is S jk_index ＝(S jk -S min )/(S max -S min )。
And respectively carrying out normalization processing on the negative emotion probabilities of the standing trigger words with all values not being null, so that the corrected values of all the negative emotion probabilities are distributed between 0 and 1, and then taking the negative numbers of the corrected values, so that the final corrected values of the negative emotion probability scores are distributed between-1 and 0. For each emotion probability score S' jk Its correction value S' jk_index The calculation method of (2) is as follows, wherein S' min Is the minimum of all negative emotion probability scores, S' max Is the maximum value in all negative emotion probability scores, and the normalization processing formula is as follows: s'. j'k_index ＝-(S’ jk -S’ min )/(S’ max -S’ min )。
In one embodiment, according to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the single user corresponding to the standing trigger word corresponding to the single user, and calculating the average emotion probability of the standing trigger word in the community formed by all the users, including:
averaging emotion probability correction values of standing trigger words triggered by standing corresponding to each user, and calculating average emotion probability of the standing trigger words in the community;
and sorting according to the average emotion probability of each standing trigger word.
Specifically, for each position trigger word k in the position trigger word library, the emotion probability score correction value S for all the user texts in the community for which the position trigger word is used jk_index And (5) averaging to obtain the average emotion probability of the standing trigger words in the communities. And then, according to the community overall average emotion probability of each standing trigger word, arranging the standing trigger words in a sequence from high to low.
In order to more clearly illustrate the specific implementation of the method of the present invention, the following description is given by way of specific example:
1. mapping examples to build user and standpoint trigger word stock
(1) Original text of the user (i.e., text information issued by the user):
each animation will convey some of the mind to children with a carefully planned scenario. The sky story line is loved by children, and christmas after growth do not have Santa Claus. While the dream is not abandoned, the animal biscuits are naturally the subject of the past cartoon, in the magical circus animal biscuits, father, which is a man, is very likely to be unable to restore the body, and is nearly crashed and is feared. The mother who consistently inherits the "firm" is surmised and has no idea in the face of the established fact that the mother is about to become immediately. The baby girl is born with natural mussels, the ' i like the Daddy … … ' that can not be restored now ' is a lovely playing partner, a recreation toy and lovely loving pet of the girl by using the eye mind to be depended on by worship, of course, there is no substitutable omnipotent hero, which allows the owner of the man to retrieve the significance of survival. Parents and children are co-growing, you do children for 10 years, I am no more than you a day, we struggle with you to go into the bar over the air
(2) In the original text, the standing trigger word is "magical circus animal biscuit", the sentence in which the word is located is taken, and n sentences (n=2 for example) are taken to form the to-be-analyzed language block.
Each animation will convey some of the mind to children with a carefully planned scenario. The sky story line is loved by children, and christmas after growth do not have Santa Claus. While the dream is not abandoned, the animal biscuits are naturally the subject of the past cartoon, in the magical circus animal biscuits, father, which is a man, is very likely to be unable to restore the body, and is nearly crashed and is feared. The mother who consistently inherits the "firm" is surmised and has no idea in the face of the established fact that the mother is about to become immediately. The baby girl is born with natural mussels, the ' i like the Daddy … … ' that can not be restored now ' is a lovely playing partner, a recreation toy and lovely loving pet of the girl by using the eye mind to be depended on by worship, of course, there is no substitutable omnipotent hero, which allows the owner of the man to retrieve the significance of survival.
(3) Calculating emotion probability of the language material block by using the emotion analysis mode shown in the foregoing to obtain:
Probability of emotion S index ＝0.979658
2. Standpoint triggering examples of data (part of) of word stock and user emotion
| Setting trigger word | Affective probability (score) of user 1 | Affective probability (score) for user 2 |
| Lunar light box for Dahui Xiyou | -0.9113407 | 0.912788 |
| Rescue redemption of xiaosheng | 0.984257 | 0.980452 |
| Bawangbai Ji | 0.98767 | 0.969589 |
| Taitannik number | 0.926384 | 0.97395 |
| The killer is not too cold | 0.986618 | 0.986399 |
| Space for stealing dream | 0.873364 | 0.966804 |
| Three fool big alarm Laiyai dock | 0.928555 | 0.983658 |
Example two
Based on the same inventive concept, this embodiment provides a user portrait creation device based on a user emotion standpoint, please refer to fig. 2, including:
the corpus acquisition module 201 is configured to acquire an independent short text corpus from user history data;
the emotion word bank construction module 202 is used for classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word bank according to the distribution condition of words in the short text corpus classification result, and calculating the original emotion probability of emotion words in the emotion word bank; wherein, the emotion word library comprises positive emotion words and negative emotion words;
the setting trigger word stock construction module 203 is configured to construct a setting trigger word stock according to an application scenario, where the setting trigger word stock includes setting trigger words that can cause a user to stand or react in emotion;
the corpus block emotion probability calculation module 204 to be analyzed is used for extracting standing trigger words contained in text information issued by a user, forming a corpus block to be analyzed according to the extracted standing trigger words, and calculating the emotion probability of the corpus block to be analyzed according to the emotion probability of the emotion words in the corpus block to be analyzed, the number of degree adverbs and the number of negatives;
The single user standing trigger word emotion probability calculation module 205 is configured to calculate, according to emotion probabilities of the to-be-analyzed speaker blocks, emotion probabilities of standing trigger words corresponding to a single user;
the average emotion probability ranking module 206 is configured to calculate average emotion probabilities of the standing trigger words in the community formed by all users according to emotion probabilities of the standing trigger words corresponding to the individual users, and rank according to the average emotion probabilities;
the user portrait construction module 207 constructs a user portrait based on the ranking of the standing trigger words in the community and the emotion probability of the standing trigger words by the individual users.
In one embodiment, emotion word library construction module 202 is specifically configured to:
classifying the obtained short text corpus according to emotion tendencies, and dividing the short text corpus into positive corpus, neutral corpus and negative corpus;
dividing the classified corpus into words and removing redundancy to obtain a corpus word library;
counting the distribution condition of each word in the corpus word library in positive corpus, neutral corpus and negative corpus;
according to the distribution condition of words, screening words related to positive and negative directions by combining chi-square verification as mark word candidates of emotion tendencies;
Screening the mark word candidates, deleting words which are not matched with the corresponding emotion tendencies, and constructing an emotion word library;
searching all original corpus corresponding to each positive emotion word, calculating the average value of the positive emotion probabilities, taking the average value as the original emotion probability of the positive emotion word, and subtracting the average value of the positive emotion probabilities from 1 for negative emotion words in an emotion word bank to obtain the original emotion probability of the negative emotion word.
In one embodiment, the to-be-analyzed corpus emotion probability calculation module 204 is specifically configured to:
forming sentences in which the extracted standing trigger words are located and n sentences before and after the sentences to form a to-be-analyzed language block, wherein n is a positive integer greater than or equal to 1;
searching positive emotion words and negative emotion words appearing in the to-be-analyzed language material block, and acquiring original emotion probability of each positive emotion word and negative emotion word;
determining a negative coefficient and a degree weight according to the number of negative words and degree adverbs in a preset range and according to each positive emotion word and each negative emotion word;
calculating an emotion probability correction value of each emotion word according to the original emotion probability, the negative coefficient and the degree weight of the emotion word;
and calculating the emotion probability of the to-be-analyzed language block according to the emotion probability correction value of the emotion words, the number of positive emotion words and the number of negative emotion words.
In one embodiment, the single user standpoint trigger word emotion probability calculation module 205 is specifically configured to:
when the standing trigger word does not appear in the data issued by the user, the emotion probability of the standing trigger word is null;
when the standing trigger word appears once in the data issued by the user, the emotion probability corresponding to the corpus where the standing trigger word appears is used as the emotion probability of the standing trigger word;
when the standing trigger word appears twice or more in the data issued by the user, taking the average value of the emotion probabilities corresponding to all corpus blocks where the standing trigger word appears as the emotion probability of the standing trigger word.
In one embodiment, the device further includes a normalization processing module, configured to calculate, according to the emotion probability of the speech block to be analyzed, the emotion probability of the position trigger word corresponding to the single user, and then:
and normalizing the emotion probability of the standing trigger words corresponding to the single user to obtain the emotion probability correction value of each standing trigger word corresponding to the single user.
In one embodiment, the average emotion probability ranking module 206 is specifically configured to:
averaging emotion probability correction values of standing trigger words triggered by standing corresponding to each user, and calculating average emotion probability of the standing trigger words in the community;
And sorting according to the average emotion probability of each standing trigger word.
Since the device described in the second embodiment of the present invention is a device used for implementing the user portrait creation method based on the emotion standpoint of the user in the first embodiment of the present invention, based on the method described in the first embodiment of the present invention, a person skilled in the art can know the specific structure and deformation of the device, and therefore, the detailed description thereof is omitted herein. All devices used in the method according to the first embodiment of the present invention are within the scope of the present invention.
Example III
Based on the same inventive concept, the application also provides a user portrait visualization method, which specifically comprises the step of performing visual display on the constructed user portrait.
In one embodiment, visually displaying a representation of a user includes:
mapping the standing trigger words to word blocks with preset shapes according to the average emotion probability;
constructing a corresponding relation between the emotion probability of a single user to the standing trigger word and the color characteristics;
and visually displaying the user portrait according to the corresponding relation between the emotion probability and the color characteristics.
In the specific implementation process, the standing trigger words can be mapped to a geometric shape body, namely a "standing word block", and according to the arrangement sequence of the standing trigger words in the community in the first embodiment, the standing word blocks are arranged to form an integral geometric pattern.
For example, a square is taken as a geometric shape body of a standing word block, 300 word banks are 300 small squares, the length of the square is 30 small squares, the width of the square is 10 small squares, and finally a regular graph is formed (for example, a rectangle with the length-width ratio of 3:1, wherein the 1 st standing trigger word is positioned on the 1 st row and the 1 st column, the 30 th standing trigger word is positioned on the 1 st row and the 30 th column, and the 300 th standing trigger word is positioned on the 10 th row and the 30 th column).
And constructing a corresponding relation between the emotion probability of the single user to the standing trigger word and the color characteristics, and carrying out visual display on the user portrait according to the constructed corresponding relation.
In the implementation process, the color gradient formula can be utilized to color each small square by the emotional response of the user according to the different color systems and intensities of the colors.
The gradient color calculating method comprises the following steps: the gradation of two colors is that the color gradation formulase:Sub>A is calculated for each of the RGB channels of A, B, respectively, with gradient=a+ (B-ase:Sub>A) x p, and for each color channel, ase:Sub>A is the value of color ase:Sub>A on that channel, and B is the value of color B on that channel. p is the percentage of the target color between AB.
For example, the red color system is selected as the representative color of the positive emotion or the positive emotion is stronger, the emotion probability score correction value S jk_index The closer to 1 the darker the color. The weaker the positive emotion, each emotion probability score S jk_index The closer to 0, the smaller the color of the square of the positive emotion, the strongest is red (255, 0) and the weakest is light gray (245,245,245), at which time the correspondence (color calculation method is) R jk ＝255+(255-245)*S jk_index ,G jk ＝255+(255-0)*S jk_index ,B jk ＝255+(255-0)*S jk_index 。
Similarly, the blue color system is selected as the representative color of the negative emotion or the opposite, the stronger the negative emotion is, the emotion probability score correction value S jk_index The closer to 1, the darker the color. The weaker the negative emotion is, the emotion probability score correction value S jk_index The closer to 0. Therefore, the color of the small square with the negative emotion is blue (0,0,255) with the strongest negative emotion and light gray (245,245,245) with the weakest negative emotion, and the corresponding relationship (color calculation method) is R jk ＝245+(245-0)*S jk_index ,G jk ＝245+(245-0)*S jk_index ,B jk ＝255+(255-245)*S jk_index 。
For a standing word block whose emotion probability is null, its color is white (255 ).
FIG. 3 is a schematic diagram showing the visualization effect of a user portrait of a single user in an embodiment of the present invention. An example diagram is an emotional response of a user to a standpoint trigger word stock, for example, a red system may be used as a representative color of a support or positive emotion, the strongest color of which is (255, 0), and the weakest RGB color is (245,245,245); the blue color is used as the representative color of the opposite or negative emotion from the standpoint, the color with the strongest value is (0,0,255), and the RGB color with the weakest value is (245,245,245); the missing color of the emotional response is white (255 ).
In addition, comparison of different user portraits and analysis of group user portraits can be performed.
Comparison of portraits for different users
Two users are selected, the emotional reactions of the two users for each standing word block are analyzed, if the emotional reactions of the two users are consistent, one color is used for representing, and if the emotional reactions of the two users are inconsistent, the other color is used for representing. And selecting the transition colors of the same color system according to the consistency degree for display. The blocks of the standing words, one of which has no emotional response or neither of which has emotional response, are represented by white. Referring specifically to fig. 4, the correspondence between the emotion probability and the color feature of the standing trigger word by the single user may be implemented in a similar manner as described above, which is not described herein again. An example graph is a comparison of the emotional responses of two users to a standpoint trigger thesaurus, e.g., the emotional responses agree to a green family, with the strongest RGB color (0,201,13) and the weakest RGB color (245,245,245); the emotional response inconsistency is a purple color system, the RGB color with the strongest value is (115,9,170), and the color with the weakest value is (245,245,245); the missing emotional response was white (255 )
Group user portrayal analysis
And for a certain user group, counting the average emotional response of each standing word block, and counting the number of positive or negative directions during counting, wherein numerical value average is not performed. For the standing word blocks with the emotional response of the user exceeding u% (u is a preset value), the percentage of the total number of users occupied by positive or negative emotion of each standing word block is counted, one color is presented when the consistency is higher, the other color is presented when the consistency is lower, and the degree of consistency of the user group views is reflected by the corresponding transition color. If the number of users who have emotional reactions to a block of standing words is less than u%, it is represented in white. Referring specifically to fig. 5, the correspondence between the emotion probability and the color feature of the standing trigger word by the single user may be implemented in a similar manner as described above, which is not described herein again.
FIG. 5 shows a comparison of the emotional responses of a user population to a standpoint trigger thesaurus, e.g., the emotional responses are consistent in a green color family, with the strongest color (0,201,13) and the weakest RGB color (245,245,245); the emotional response inconsistency is a purple color system, the color of the strongest value is (115,9,170), and the RGB color of the weakest value is (245,245,245); the missing color of the emotional response is white (255 ).
According to the method provided by the invention, the emotion word library can be constructed according to the distribution condition of words in the short text corpus classification result, the constructed emotion word library is divided into positive emotion words and negative emotion words according to emotion tendencies, emotion tendencies of new words can be better identified from text contents of users, emotion reactions in the new words can be accurately analyzed, a standing trigger word library is constructed, mapping between a single user and the standing trigger word library is then established, and emotion probability of standing trigger words corresponding to the single user is calculated; and then calculating the average emotion probability of the standing trigger words in the community formed by all the users, and sorting the average emotion probability, so that user portraits can be constructed according to the sorting result, the emotion standing of each user on the standing trigger words can be accurately realized, the viewpoint difference and emotion response characteristics of each user on common things can be rapidly and accurately known, and the technical problem that the analysis result is inaccurate in the method in the prior art is solved.
Furthermore, based on the constructed user portrait, the invention also provides a visualization method of the user portrait, which is used for carrying out visualization display on the user portrait and improving intuitiveness.
Further, according to the color gradient formula, the corresponding relation between the emotion probability of the single user to the standing trigger words and the color characteristics is constructed, and then the user portrait is visually displayed according to the corresponding relation between the emotion probability and the color characteristics, so that the display effect can be improved.
Example IV
Referring to fig. 6, based on the same inventive concept, the present application also provides a computer-readable storage medium 300, on which a computer program 311 is stored, which when executed implements the method as described in embodiment one.
Since the computer readable storage medium described in the third embodiment of the present invention is a computer device used for implementing the user portrait creation method based on the emotion of the user in the first embodiment of the present invention, based on the method described in the first embodiment of the present invention, a person skilled in the art can understand the specific structure and the modification of the computer readable storage medium, and therefore, the details are not repeated here. All computer readable storage media used in the method of the first embodiment of the present invention are within the scope of the present invention.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the embodiments of the present invention without departing from the spirit or scope of the embodiments of the invention. Thus, if such modifications and variations of the embodiments of the present invention fall within the scope of the claims and the equivalents thereof, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A user portrayal creation method based on a user emotion standpoint, comprising:
acquiring independent short text corpus from user history data;
classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word library according to the distribution condition of words in the short text corpus classification result, and calculating the original emotion probability of emotion words in the emotion word library; wherein, the emotion word library comprises positive emotion words and negative emotion words;
constructing a standing triggering word library according to an application scene, wherein the standing triggering word library comprises standing triggering words which can cause standing or emotional response of a user;
extracting standing trigger words contained in text information issued by a user, forming a to-be-analyzed text block according to the extracted standing trigger words, and calculating the emotion probability of the to-be-analyzed text block according to the emotion probability of emotion words in the to-be-analyzed text block, the number of degree adverbs and the number of negative words;
according to the emotion probability of the to-be-analyzed language block, calculating the emotion probability of the standing trigger word corresponding to the single user;
calculating the average emotion probability of the standing trigger words in the community formed by all the users according to the emotion probability of the standing trigger words corresponding to the single user, and sorting according to the average emotion probability;
And constructing user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words.
2. The method of claim 1, wherein classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word bank according to word distribution in the short text corpus classification result, and calculating emotion probabilities of emotion words in the emotion word bank, comprising:
classifying the obtained short text corpus according to emotion tendencies, and dividing the short text corpus into positive corpus, neutral corpus and negative corpus;
dividing the classified corpus into words and removing redundancy to obtain a corpus word library;
counting the distribution condition of each word in the corpus word library in positive corpus, neutral corpus and negative corpus;
according to the distribution condition of words, screening words related to positive and negative directions by combining chi-square verification as mark word candidates of emotion tendencies;
screening the mark word candidates, deleting words which are not matched with the corresponding emotion tendencies, and constructing an emotion word library;
searching all original corpus corresponding to each positive emotion word, calculating the average value of the positive emotion probabilities, taking the average value as the original emotion probability of the positive emotion word, and subtracting the average value of the positive emotion probabilities from 1 for negative emotion words in an emotion word bank to obtain the original emotion probability of the negative emotion word.
3. The method of claim 1, wherein composing the blocks of speech to be analyzed based on the extracted position trigger words, and then calculating the emotion probability of the blocks of speech to be analyzed based on the number of emotion words, degree adverbs and negatives in the blocks of speech to be analyzed, comprises:
forming sentences in which the extracted standing trigger words are located and n sentences before and after the sentences to form a to-be-analyzed language block, wherein n is a positive integer greater than or equal to 1;
searching positive emotion words and negative emotion words appearing in the to-be-analyzed language material block, and acquiring original emotion probability of each positive emotion word and negative emotion word;
determining a negative coefficient and a degree weight according to the number of negative words and degree adverbs in a preset range and according to each positive emotion word and each negative emotion word;
calculating an emotion probability correction value of each emotion word according to the original emotion probability, the negative coefficient and the degree weight of the emotion word;
and calculating the emotion probability of the to-be-analyzed language block according to the emotion probability correction value of the emotion words, the number of positive emotion words and the number of negative emotion words.
4. The method of claim 1, wherein calculating the emotion probabilities of the standpoint trigger words corresponding to the individual users based on the emotion probabilities of the chunks of speech to be analyzed comprises:
When the standing trigger word does not appear in the data issued by the user, the emotion probability of the standing trigger word is null;
when the standing trigger word appears once in the data issued by the user, the emotion probability corresponding to the corpus where the standing trigger word appears is used as the emotion probability of the standing trigger word;
when the standing trigger word appears twice or more in the data issued by the user, taking the average value of the emotion probabilities corresponding to all corpus blocks where the standing trigger word appears as the emotion probability of the standing trigger word.
5. The method of claim 1, wherein after calculating the emotion probabilities of the standpoint trigger words corresponding to the individual users based on the emotion probabilities of the chunks of speech to be analyzed, the method further comprises:
and normalizing the emotion probability of the standing trigger words corresponding to the single user to obtain the emotion probability correction value of each standing trigger word corresponding to the single user.
6. The method of claim 5, wherein calculating the average emotion probabilities of the context trigger words within the community of all users based on the emotion probabilities of the context trigger words corresponding to the individual users comprises:
averaging emotion probability correction values of standing trigger words triggered by standing corresponding to each user, and calculating average emotion probability of the standing trigger words in the community;
And sorting according to the average emotion probability of each standing trigger word.
7. A user portrayal creation apparatus based on a user emotion standpoint, comprising:
the corpus acquisition module is used for acquiring independent short text corpus from the user history data;
the emotion word bank construction module is used for classifying the obtained short text corpus according to emotion tendencies, constructing an emotion word bank according to the distribution condition of words in the short text corpus classification result and calculating the original emotion probability of emotion words in the emotion word bank; wherein, the emotion word library comprises positive emotion words and negative emotion words;
the establishment triggering word stock construction module is used for constructing an establishment triggering word stock according to an application scene, wherein the establishment triggering word stock comprises establishment triggering words capable of causing establishment or emotional response of a user;
the corpus block emotion probability calculation module is used for extracting standing trigger words contained in text information issued by a user, forming a corpus block to be analyzed according to the extracted standing trigger words, and calculating the emotion probability of the corpus block to be analyzed according to the emotion probability of the emotion words in the corpus block to be analyzed, the number of degree auxiliary words and the number of negative words;
The single user standing trigger word emotion probability calculation module is used for calculating the emotion probability of the standing trigger word corresponding to the single user according to the emotion probability of the to-be-analyzed language block;
the average emotion probability ordering module is used for calculating the average emotion probability of the standing trigger words in the community formed by all the users according to the emotion probability of the standing trigger words corresponding to the single user, and ordering according to the average emotion probability;
and the user portrait construction module constructs user portraits according to the ordering condition of the standing trigger words in the communities and the emotion probability of the single user to the standing trigger words.
8. A method for visualizing a user representation, comprising: a visual display of a user representation constructed by the method of any one of claims 1 to 6.
9. The method of claim 8, wherein visually displaying the representation of the user comprises:
mapping the standing trigger words to word blocks with preset shapes according to the average emotion probability;
constructing a corresponding relation between the emotion probability of a single user to the standing trigger word and the color characteristics;
and visually displaying the user portrait according to the corresponding relation between the emotion probability and the color characteristics.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when executed, implements the method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910961379.5A CN112651237B (en) | 2019-10-11 | 2019-10-11 | User portrait establishing method and device based on user emotion standpoint and user portrait visualization method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910961379.5A CN112651237B (en) | 2019-10-11 | 2019-10-11 | User portrait establishing method and device based on user emotion standpoint and user portrait visualization method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112651237A CN112651237A (en) | 2021-04-13 |
| CN112651237B true CN112651237B (en) | 2024-03-19 |
Family
ID=75343004
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910961379.5A Active CN112651237B (en) | 2019-10-11 | 2019-10-11 | User portrait establishing method and device based on user emotion standpoint and user portrait visualization method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112651237B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114140814A (en) * | 2022-02-07 | 2022-03-04 | 北京无疆脑智科技有限公司 | Emotion recognition capability training method and device and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013222231A (en) * | 2012-04-13 | 2013-10-28 | Nec Corp | Emotion sharing communication facilitating system, emotion sharing communication facilitating method, and program |
| JP2017120634A (en) * | 2015-12-28 | 2017-07-06 | 株式会社リコー | Analytical method and device for sentimental word polarity |
| CN109767787A (en) * | 2019-01-28 | 2019-05-17 | 腾讯科技（深圳）有限公司 | Emotion identification method, equipment and readable storage medium storing program for executing |
| CN109994102A (en) * | 2019-04-16 | 2019-07-09 | 上海航动科技有限公司 | A kind of outer paging system of intelligence based on Emotion identification |
-
2019
- 2019-10-11 CN CN201910961379.5A patent/CN112651237B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013222231A (en) * | 2012-04-13 | 2013-10-28 | Nec Corp | Emotion sharing communication facilitating system, emotion sharing communication facilitating method, and program |
| JP2017120634A (en) * | 2015-12-28 | 2017-07-06 | 株式会社リコー | Analytical method and device for sentimental word polarity |
| CN109767787A (en) * | 2019-01-28 | 2019-05-17 | 腾讯科技（深圳）有限公司 | Emotion identification method, equipment and readable storage medium storing program for executing |
| CN109994102A (en) * | 2019-04-16 | 2019-07-09 | 上海航动科技有限公司 | A kind of outer paging system of intelligence based on Emotion identification |
Non-Patent Citations (2)
| Title |
|---|
| 索晓阳 ; 王伟 ; .基于社交网络数据的用户群体画像构建方法研究.网络空间安全.2019,(09),全文. * |
| 蒋盛益 ; 黄卫坚 ; 蔡茂丽 ; 王连喜 ; .面向微博的社会情绪词典构建及情绪分析方法研究.中文信息学报.2015,(06),全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112651237A (en) | 2021-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111415740B (en) | Method and device for processing inquiry information, storage medium and computer equipment | |
| CN106227722B (en) | A kind of extraction method based on listed company's bulletin abstract | |
| CN110188351B (en) | Sentence smoothness and syntax scoring model training method and device | |
| US9805020B2 (en) | In-context access of stored declarative knowledge using natural language expression | |
| US20170364773A1 (en) | Training a classifier algorithm used for automatically generating tags to be applied to images | |
| US8972445B2 (en) | Systems and methods for storage of declarative knowledge accessible by natural language in a computer capable of appropriately responding | |
| CN109829166B (en) | People and host customer opinion mining method based on character-level convolutional neural network | |
| CN110675944A (en) | Triage method and device, computer equipment and medium | |
| CN109840279A (en) | File classification method based on convolution loop neural network | |
| CN109902912B (en) | Personalized image aesthetic evaluation method based on character features | |
| CN106528676B (en) | Entity Semantics search processing method and device based on artificial intelligence | |
| JP2006293767A (en) | Sentence categorizing device, sentence categorizing method, and categorization dictionary creating device | |
| CN109710744A (en) | A kind of data matching method, device, equipment and storage medium | |
| WO2021017300A1 (en) | Question generation method and apparatus, computer device, and storage medium | |
| CN110083829A (en) | Feeling polarities analysis method and relevant apparatus | |
| Korsgaard et al. | Creating user stereotypes for persona development from qualitative data through semi-automatic subspace clustering | |
| CN109325115A (en) | A kind of role analysis method and analysis system | |
| KR20200010672A (en) | Smart merchandise searching method and system using deep learning | |
| CN108363699A (en) | A kind of netizen's school work mood analysis method based on Baidu's mhkc | |
| CN106886565A (en) | A kind of basic house type auto-polymerization method | |
| CN110781300B (en) | Tourism resource culture characteristic scoring algorithm based on Baidu encyclopedia knowledge graph | |
| CN112651237B (en) | User portrait establishing method and device based on user emotion standpoint and user portrait visualization method | |
| CN114416929A (en) | Sample generation method, device, equipment and storage medium of entity recall model | |
| CN114154839A (en) | Course recommendation method based on online education platform data | |
| CN112906400A (en) | Automatic subjective question scoring method based on domain ontology and semantic dependency tree |
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 |