CN109829154A - Semantic-based personality prediction technique, user equipment, storage medium and device - Google Patents
Semantic-based personality prediction technique, user equipment, storage medium and device Download PDFInfo
- Publication number
- CN109829154A CN109829154A CN201910042095.6A CN201910042095A CN109829154A CN 109829154 A CN109829154 A CN 109829154A CN 201910042095 A CN201910042095 A CN 201910042095A CN 109829154 A CN109829154 A CN 109829154A
- Authority
- CN
- China
- Prior art keywords
- vector
- semantic
- personality
- text
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention discloses semantic-based personality prediction technique, user equipment, storage medium and devices.The text feature word respectively to be operated in text to be predicted is first obtained in the present invention, is calculated semantic weight corresponding with text feature word to be operated, the text vector being made of semantic weight is clustered to obtain Semantic Clustering vector；Distributed expression processing is carried out to obtain term vector to text to be predicted again, term vector is trained based on default convolutional neural networks to obtain neural predicted vector；Semantic Clustering vector and neural predicted vector are spliced, to obtain vector to be entered, carry out the prediction of user's personality by presetting classifier according to vector to be entered.It is apparent that having combined the prediction result of cluster result and default convolutional neural networks based on context semanteme in the present invention, Clustering Effect is improved, the accuracy of personality prediction is improved, solves the lower technical problem of personality predictablity rate.
Description
Technical field
The present invention relates to personality electric powder prediction more particularly to semantic-based personality prediction technique, user equipment, deposit
Storage media and device.
Background technique
Five-factor model personality theory on psychological field has preferable application value, can be applied to marketing scene, society
It hands under the different types of scene such as network scenarios and personnel management scene, the theoretical five kinds of signified personalities of five-factor model personality are respectively
Open personality, sense of responsibility personality, extroversion personality, pleasant personality and neurotic personality.
For example, if five-factor model personality theory is applied under marketing scene, it can be by predicting the personal traits of user
Carry out the inner link between investigative analysis user personality and product, so as to more accurately recommend its favorite production for user
Product, for example, the recommendation of commercial product recommending, music and love and marriage recommendation etc..
If five-factor model personality theory is applied under social networks scene, can be analyzed by predicting the personal traits of user
The behavioral characteristic of the network user, and then the influence power of user can be explored, analyze the group of Web Community and to the network user
Improper behavior and bad speech be monitored and administer.
If five-factor model personality theory is applied under personnel management scene, it is contemplated that the on-job industry selection of the different personalities of people and people
Thing selects aspect, and there is close relevances, for example, the extroversion and open personality in research discovery five-factor model personality are duties
Two key factors of industry psychology and Industrial Psychology, and have the personality of duty confidence and personnel selection even more closely related.So
In terms of personal job hunting and company selects aspect, and the personality trait of a clear people is for personal and company long-term interest
It is very crucial.
It is apparent that personality predicts application value with higher.But for social networks scene, there is a variety of
The prediction mode of personality prediction, for example, the text occurred in social networks, head portrait, expression and user response mode can be integrated
Etc. information carry out personality prediction, still, the accuracy rate of the prediction result of most personality prediction is lower.As it can be seen that personality is pre-
The survey mode technical problem lower there is predictablity rate.
Above content is only used to facilitate the understanding of the technical scheme, and is not represented and is recognized that above content is existing skill
Art.
Summary of the invention
The main purpose of the present invention is to provide semantic-based personality prediction technique, user equipment, storage medium and dresses
It sets, it is intended to solve the lower technical problem of personality predictablity rate.
To achieve the above object, the present invention provides a kind of semantic-based personality prediction technique, the semantic-based people
Lattice prediction technique the following steps are included:
To in text set to be predicted text to be predicted carry out word segmentation processing, with obtain in the text to be predicted respectively to
Operate text feature word；
Calculate semantic weight corresponding with the text feature word to be operated；
The text vector being made of the semantic weight is clustered, to obtain Semantic Clustering vector；
Distributed expression processing is carried out to the text to be predicted, to obtain term vector；
The term vector is trained based on default convolutional neural networks, to obtain neural predicted vector；
Institute's semantic cluster vector and the neural predicted vector are spliced, to obtain vector to be entered；
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain personality prediction knot
Fruit.
Preferably, calculating semantic weight corresponding with the text feature word to be operated, comprising:
Default frequency of occurrence of the co-occurrence Feature Words in the text set to be predicted is counted, the default co-occurrence Feature Words are
The two text feature words to be operated occurred in a text to be predicted；
According to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence.
Preferably, described that the text vector being made of the semantic weight is clustered, to obtain Semantic Clustering vector,
Include:
According to the semantic weight construction feature word matrix；
Singular value decomposition is carried out to the Feature Words matrix, to obtain semantic matrix；
Text vector in the semantic matrix is clustered, to obtain Semantic Clustering vector.
Preferably, described that the term vector is trained based on default convolutional neural networks, to obtain the pre- direction finding of nerve
Amount, comprising:
Convolution sum pond is carried out to the term vector based on default convolutional neural networks, to obtain feature vector；
Default word frequency vector is spliced with described eigenvector, to obtain splicing vector；
The splicing vector is trained, to obtain prediction probability value, and prediction probability value composition nerve is pre-
Direction finding amount.
Preferably, the prediction for carrying out user's personality by presetting classifier according to the vector to be entered, to obtain
Personality prediction result, comprising:
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain and the user people
The corresponding personality probability of lattice；
When the personality probability is more than or equal to predetermined probabilities threshold value, the personality probability of the predetermined probabilities threshold value will be greater than
Corresponding user's personality is as personality prediction result.
Preferably, the prediction for carrying out user's personality by presetting classifier according to the vector to be entered, to obtain
Before personality prediction result, the semantic-based personality prediction technique is further comprising the steps of:
It obtains to training vector；
Based on Adaboost algorithm it is determining with described to the corresponding initial weight of training vector；
It is trained to described to training vector based on the initial weight, to obtain the first Weak Classifier；
Determine the first error rate corresponding with first Weak Classifier；
When first error rate is less than or equal to default error rate threshold, the Weak Classifier of acquisition is combined, with
Combination obtains strong classifier, and the strong classifier is regarded as default classifier.
Preferably, described semantic-based after the determination the first error rate corresponding with first Weak Classifier
Personality prediction technique is further comprising the steps of:
When first error rate is greater than the default error rate threshold, the initial weight is adjusted, to obtain
Obtain the second weight；
It is trained to described to training vector based on second weight, to obtain the second Weak Classifier；
Determine the second error rate corresponding with second Weak Classifier；
When second error rate is less than or equal to the default error rate threshold, the Weak Classifier of acquisition is subjected to group
It closes, strong classifier is obtained with combination, and the strong classifier is regarded as into default classifier.
In addition, to achieve the above object, the present invention also proposes a kind of user equipment, the user equipment include memory,
Processor and the semantic-based personality Prediction program that is stored on the memory and can run on the processor, it is described
Semantic-based personality Prediction program is arranged for carrying out the step of semantic-based personality prediction technique as described above.
In addition, to achieve the above object, the present invention also proposes a kind of storage medium, it is stored with and is based on the storage medium
Semantic personality Prediction program, the semantic-based personality Prediction program realize base as described above when being executed by processor
In the semantic personality prediction technique the step of.
In addition, to achieve the above object, the present invention also proposes a kind of semantic-based personality prediction meanss, described to be based on language
Justice personality prediction meanss include:
Word segmentation processing module, for carrying out word segmentation processing to the text to be predicted in text set to be predicted, described in obtaining
Text feature word respectively to be operated in text to be predicted；
Semantic weight computing module, for calculating semantic weight corresponding with the text feature word to be operated；
Cluster module, for being clustered to the text vector being made of the semantic weight, with obtain Semantic Clustering to
Amount；
Distributed processing modules, for carrying out distributed expression processing to the text to be predicted, to obtain term vector；
Neural network module, for being trained based on default convolutional neural networks to the term vector, to obtain nerve
Predicted vector；
Vector splicing module, for splicing to institute's semantic cluster vector and the neural predicted vector, to obtain
Vector to be entered；
Personality prediction module, for the prediction of user's personality to be carried out by presetting classifier according to the vector to be entered,
To obtain personality prediction result.
The text feature word respectively to be operated in text to be predicted is first obtained in the present invention, is calculated and text feature word to be operated
Corresponding semantic weight clusters the text vector being made of semantic weight to obtain Semantic Clustering vector；It treats again pre-
It surveys text and carries out distributed expression processing to obtain term vector, term vector is trained to obtain based on default convolutional neural networks
Obtain neural predicted vector；Semantic Clustering vector and neural predicted vector are spliced, to obtain vector to be entered, according to defeated
Incoming vector carries out the prediction of user's personality by presetting classifier.It is apparent that having combined in the present invention based on context semanteme
The prediction result of cluster result and default convolutional neural networks, improves Clustering Effect, improves the accuracy of personality prediction,
Solves the lower technical problem of personality predictablity rate.
Detailed description of the invention
Fig. 1 is the user device architecture schematic diagram for the hardware running environment that the embodiment of the present invention is related to；
Fig. 2 is that the present invention is based on the flow diagrams of semantic personality prediction technique first embodiment；
Fig. 3 is that the present invention is based on the flow diagrams of semantic personality prediction technique second embodiment；
Fig. 4 is that the present invention is based on the flow diagrams of semantic personality prediction technique 3rd embodiment；
Fig. 5 is that the present invention is based on the structural block diagrams of semantic personality prediction meanss first embodiment.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Referring to Fig.1, Fig. 1 is the user device architecture schematic diagram for the hardware running environment that the embodiment of the present invention is related to.
As shown in Figure 1, the user equipment may include: processor 1001, such as CPU, communication bus 1002, user interface
1003, network interface 1004, memory 1005.Wherein, communication bus 1002 is for realizing the connection communication between these components.
User interface 1003 may include display screen (Display), optional user interface 1003 can also include standard wireline interface,
Wireless interface, the wireline interface for user interface 1003 can be USB interface in the present invention.Network interface 1004 optionally may be used
To include standard wireline interface and wireless interface (such as WI-FI interface).Memory 1005 can be high speed RAM memory, can also
To be stable memory (non-volatile memory), such as magnetic disk storage.Memory 1005 optionally can also be
Independently of the storage device of aforementioned processor 1001.
It will be understood by those skilled in the art that structure shown in Fig. 1 does not constitute the restriction to user equipment, can wrap
It includes than illustrating more or fewer components, perhaps combines certain components or different component layouts.
As shown in Figure 1, as may include that operating system, network are logical in a kind of memory 1005 of computer storage medium
Believe module, Subscriber Interface Module SIM and semantic-based personality Prediction program.
In user equipment shown in Fig. 1, network interface 1004 is mainly used for connecting background server, takes with the backstage
Business device carries out data communication；User interface 1003 is mainly used for connecting peripheral hardware；The user equipment is called by processor 1001
The semantic-based personality Prediction program stored in memory 1005, and execute following operation:
To in text set to be predicted text to be predicted carry out word segmentation processing, with obtain in the text to be predicted respectively to
Operate text feature word；
Calculate semantic weight corresponding with the text feature word to be operated；
The text vector being made of the semantic weight is clustered, to obtain Semantic Clustering vector；
Distributed expression processing is carried out to the text to be predicted, to obtain term vector；
The term vector is trained based on default convolutional neural networks, to obtain neural predicted vector；
Institute's semantic cluster vector and the neural predicted vector are spliced, to obtain vector to be entered；
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain personality prediction knot
Fruit.
Further, processor 1001 can call the semantic-based personality Prediction program stored in memory 1005,
Also execute following operation:
Default frequency of occurrence of the co-occurrence Feature Words in the text set to be predicted is counted, the default co-occurrence Feature Words are
The two text feature words to be operated occurred in a text to be predicted；
According to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence.
Further, processor 1001 can call the semantic-based personality Prediction program stored in memory 1005,
Also execute following operation:
According to the semantic weight construction feature word matrix；
Singular value decomposition is carried out to the Feature Words matrix, to obtain semantic matrix；
Text vector in the semantic matrix is clustered, to obtain Semantic Clustering vector.
Further, processor 1001 can call the semantic-based personality Prediction program stored in memory 1005,
Also execute following operation:
Convolution sum pond is carried out to the term vector based on default convolutional neural networks, to obtain feature vector；
Default word frequency vector is spliced with described eigenvector, to obtain splicing vector；
The splicing vector is trained, to obtain prediction probability value, and prediction probability value composition nerve is pre-
Direction finding amount.
Further, processor 1001 can call the semantic-based personality Prediction program stored in memory 1005,
Also execute following operation:
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain and the user people
The corresponding personality probability of lattice；
When the personality probability is more than or equal to predetermined probabilities threshold value, the personality probability of the predetermined probabilities threshold value will be greater than
Corresponding user's personality is as personality prediction result.
Further, processor 1001 can call the semantic-based personality Prediction program stored in memory 1005,
Also execute following operation:
It obtains to training vector；
Based on Adaboost algorithm it is determining with described to the corresponding initial weight of training vector；
It is trained to described to training vector based on the initial weight, to obtain the first Weak Classifier；
Determine the first error rate corresponding with first Weak Classifier；
When first error rate is less than or equal to default error rate threshold, the Weak Classifier of acquisition is combined, with
Combination obtains strong classifier, and the strong classifier is regarded as default classifier.
Further, processor 1001 can call the semantic-based personality Prediction program stored in memory 1005,
Also execute following operation:
When first error rate is greater than the default error rate threshold, the initial weight is adjusted, to obtain
Obtain the second weight；
It is trained to described to training vector based on second weight, to obtain the second Weak Classifier；
Determine the second error rate corresponding with second Weak Classifier；
When second error rate is less than or equal to the default error rate threshold, the Weak Classifier of acquisition is subjected to group
It closes, strong classifier is obtained with combination, and the strong classifier is regarded as into default classifier.
The text feature word respectively to be operated in text to be predicted is first obtained in the present embodiment, is calculated and text feature to be operated
The corresponding semantic weight of word clusters the text vector being made of semantic weight to obtain Semantic Clustering vector；It treats again
Predict text carry out it is distributed indicate processing to obtain term vector, based on default convolutional neural networks to term vector be trained with
Obtain neural predicted vector；Semantic Clustering vector and neural predicted vector are spliced, to obtain vector to be entered, according to
Input vector carries out the prediction of user's personality by presetting classifier.It is apparent that having combined in the present embodiment based on context language
The cluster result of justice and the prediction result of default convolutional neural networks, improve Clustering Effect, improve the standard of personality prediction
True property solves the lower technical problem of personality predictablity rate.
Based on above-mentioned hardware configuration, propose that the present invention is based on the embodiments of semantic personality prediction technique.
It is that the present invention is based on the flow diagrams of semantic personality prediction technique first embodiment referring to Fig. 2, Fig. 2.
In the first embodiment, the semantic-based personality prediction technique the following steps are included:
Step S10: word segmentation processing is carried out to the text to be predicted in text set to be predicted, to obtain the text to be predicted
In text feature word respectively to be operated.
It is understood that can first be obtained from social networks big to realize that the personality under social networks scene is predicted
The text to be predicted of amount, text set to be predicted are denoted as the set of a large amount of text to be predicted.Text to be predicted can be user A in society
The text information generated in network is handed over, the personality prediction of user A can be carried out based on the text to be predicted.
In the concrete realization, first the text to be predicted can be pre-processed, by text dividing to be predicted at single one by one
Only word.
Step S20: semantic weight corresponding with the text feature word to be operated is calculated.
It should be understood that semantic weight corresponding with the text feature word to be operated being syncopated as, semantic weight can be calculated
For characterizing the correlation degree between context semanteme.It is apparent that by introducing the semantic weight with reference to context semanteme, it can
To improve the extraction precision for semantic feature.
Step S30: clustering the text vector being made of the semantic weight, to obtain Semantic Clustering vector.
It is understood that the text vector being made of semantic weight can be clustered based on K-means clustering algorithm,
To obtain Semantic Clustering vector.Wherein, it is contemplated that each text feature word is there is corresponding semantic weight, and text feature
Word is combined into short, the semantic weight of each text feature word of a word can be constituted one-dimensional text vector.In addition,
K-means clustering algorithm belongs to a kind of typical hard clustering algorithm.
Step S40: distributed expression processing is carried out to the text to be predicted, to obtain term vector.
It is understood that step S10-S30 is by handling text to be predicted, the Semantic Clustering handled to
Amount can characterize the correlation degree between context semanteme.But other than the cluster for context semanteme, it can also pass through step
Rapid S40-50 carries out the prediction of convolutional neural networks, by with reference to the relevance and convolutional Neural between context semanteme
The prediction result of network can preferably improve the accuracy of prediction result.
In the concrete realization, it in order to carry out personality prediction based on convolutional neural networks, can first text to be predicted be distributed
Formula expression processing, textual form to be predicted, which is turned to vector, to be indicated, to obtain term vector.
Step S50: the term vector is trained based on default convolutional neural networks, to obtain neural predicted vector.
It should be understood that term vector can be inputted to the input layer of default convolutional neural networks, led to after obtaining term vector
Training is crossed to realize the neural predicted vector of final acquisition.
Step S60: splicing institute's semantic cluster vector and the neural predicted vector, with obtain it is to be entered to
Amount.
It is understood that the Semantic Clustering vector that context Semantic Clustering obtains may tie up output vector for m, volume is preset
The neural predicted vector that product neural network prediction obtains may tie up output vector for n, and m and n are positive integer, can combination semantic cluster
Vector and neural predicted vector carry out integrated prediction.
Step S70: the prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain personality
Prediction result.
It should be understood that the vector that will be tieed up by the vector to be entered that concatenation obtains for m+n, and the vector is returned
Input vector after one change as integrated prediction.
In the concrete realization, if using the foundation that five-factor model personality theory is predicted as personality, can input into default classifier should
Vector to be entered carries out the prediction of the personality based on five-factor model personality by default classifier, to export personality prediction result.Wherein,
Personality prediction result may be the personality or multinomial personality in five-factor model personality theory.
The text feature word respectively to be operated in text to be predicted is first obtained in the present embodiment, is calculated and text feature to be operated
The corresponding semantic weight of word clusters the text vector being made of semantic weight to obtain Semantic Clustering vector；It treats again
Predict text carry out it is distributed indicate processing to obtain term vector, based on default convolutional neural networks to term vector be trained with
Obtain neural predicted vector；Semantic Clustering vector and neural predicted vector are spliced, to obtain vector to be entered, according to
Input vector carries out the prediction of user's personality by presetting classifier.It is apparent that having combined in the present embodiment based on context language
The cluster result of justice and the prediction result of default convolutional neural networks, improve Clustering Effect, improve the standard of personality prediction
True property solves the lower technical problem of personality predictablity rate.
It is to be based on the present invention is based on the flow diagram of semantic personality prediction technique second embodiment referring to Fig. 3, Fig. 3
Above-mentioned first embodiment shown in Fig. 2 proposes that the present invention is based on the second embodiments of semantic personality prediction technique.
In second embodiment, the text to be predicted in text set to be predicted carries out word segmentation processing, described in obtaining
Text feature word respectively to be operated in text to be predicted, comprising:
Word segmentation processing is carried out to text to be predicted, to obtain each text feature word in the text to be predicted；
Default stop words is removed from the text feature word, and the text feature word for removing the default stop words is recognized
It is set to text feature word to be operated.
It in the concrete realization, can be from text feature in order to improve the efficiency of prediction process and the accuracy of prediction result
Weed out the default stop words there is no significant meaning in word, default stop words may include " ", " " and " I " etc..
Further, the step S20, comprising:
Step S201: counting default frequency of occurrence of the co-occurrence Feature Words in the text set to be predicted, described default total
Existing Feature Words are the two text feature words to be operated occurred in a text to be predicted.
It is understood that in order to determine semantic weight, the statistics available co-occurrence word pair occurred in a text to be predicted is total
Existing word to appearing in a text for two Feature Words and simultaneously.If two text feature word fw to be operatedaAnd fwbIt can simultaneously
Occur in a text to be predicted, and appears in multiple texts to be predicted in text set to be predicted, if occur simultaneously
Number is not less than 2, then the two text feature words to be operated can be regarded as to a co-occurrence word pair, and two words are respectively default total
Existing Feature Words.
It should be understood that text set to be predicted can be traversed and count default co-occurrence Feature Words in text set to be predicted
Frequency of occurrence can be known as co-occurrence degree, be calculated as FW by frequency of occurrenceab.Meanwhile it can be by FWabIt is included in co-occurrence degree set Co (FWab)。
Step S202: according to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence.
It is understood that after obtaining co-occurrence degree, it can be corresponding with the default co-occurrence Feature Words according to the determination of co-occurrence degree
Semantic weight, semantic weight can be denoted as to Semantic Weight of Feature Word (SWFW).
In the concrete realization, it can be preset according to the frequency of occurrence is determining with described under default semantic weight calculation formula
The corresponding semantic weight of co-occurrence Feature Words, preset semantic weight calculation formula specifically,
Wherein, SWFWaRefer to default co-occurrence Feature Words fwaSemantic weight, TFIDFbIt is default co-occurrence Feature Words fwb's
Word frequency-inverted file frequency (term frequency-inverse document frequency, TF-IDF) weight, Co
(FWab) refer to fwaAnd fwbCo-occurrence degree, M be in text set to be predicted vocabulary sum, N is in text set to be predicted to pre-
Survey the text number of text.
It should be understood that TFIDF weight is to assess a Feature Words for the weight of a text set or a text
Want degree.So TFIDF weight corresponding with the text feature word to be operated can be calculated, then calculated in default semantic weight
According to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence and the TFIDF weight under formula.
Further, the step S30, comprising:
Step S301: according to the semantic weight construction feature word matrix.
It is understood that obtaining each after operating the semantic weight of text feature word, can by semantic weight come
Construction feature word matrix, Feature Words matrix is by the relationship between characteristic feature word and text to be predicted.
In the concrete realization, it can be weighted summation based on the TFIDF weight and the semantic weight, to obtain newly
Final weight of the semantic weight as text feature word to be operated, and according to new semantic weight construction feature word matrix.
Step S302: singular value decomposition is carried out to the Feature Words matrix, to obtain semantic matrix.
It should be understood that applicable singular value decomposition mode carries out Feature Words matrix after obtaining Feature Words matrix
Matrix decomposition, and retain m singular value of preceding preset quantity, to be based on the corresponding text feature word structure to be operated of m singular value
Semantic matrix is built out, semantic matrix can represent the text vector spatial model based on context semanteme.Wherein, m is positive integer.
Step S303: clustering the text vector in the semantic matrix, to obtain Semantic Clustering vector.
It is understood that after obtaining semantic matrix, it can be based on K-means clustering algorithm to the text in semantic matrix
Vector is clustered, to obtain cluster result.And count the summation of number of objects corresponding with default cluster in the cluster result
Number, it is contemplated that the quantity of the cluster centre in cluster process is m, then can indicate in vector form m total sum numbers, to realize
Expression for user A.So Semantic Clustering vector can be denoted as the total value of vectorization.
It should be understood that can be effectively improved short by carrying out cluster operation under the premise of combining context semantic
The insufficient defect of the feature of text and higher-dimension sparsity, to optimize Clustering Effect.
Further, described that the term vector is trained based on default convolutional neural networks, to obtain nerve prediction
Vector, comprising:
Convolution sum pond is carried out to the term vector based on default convolutional neural networks, to obtain feature vector；
Default word frequency vector is spliced with described eigenvector, to obtain splicing vector；
The splicing vector is trained, to obtain prediction probability value, and prediction probability value composition nerve is pre-
Direction finding amount.
It is understood that can also introduce default convolutional Neural other than with reference to the cluster result based on context semanteme
The prediction result of network (Convolutional Neural Networks, CNN) is as reference quantity.
It should be understood that personality prediction can be carried out based on Text-CNN, specifically, the input layer of Text-CNN will be
The convolutional layer of sentence matrix, Text-CNN will be made of 3 convolution kernels.So can be based on the Text-CNN that this kind constructs to institute
Predicate vector is handled, with output nerve predicted vector.
In the concrete realization, sentence length is first set as preset value, and the sentence based on the preset value is long constitutes term vector
For sentence matrix, then this sentence matrix done into convolution operation, the one-dimensional convolution of multichannel for being respectively 1,2,3 by convolution kernel size
Layer, convolution obtain 3 characteristic layers.Again by maximum pond to each characteristic layer dimensionality reduction, and the result that dimensionality reduction is obtained is regarded as
Feature vector.After obtaining all feature vectors, word frequency vector can be preset and spliced with all feature vectors, and will spelled
Splicing vector after connecing continues to train as the input of next stage.Wherein, word frequency vector concretely LIWC is preset
(the Linguistic Inquiry Word Count, language inquiry and number of words) vector.It then, can be in the complete of Text-CNN
The splicing vector for tieing up the n of acquisition in articulamentum assigns soft-max classifier, so that soft-max classifier exports n difference
The class probability value of classification, and n different classes of class probability values are combined into n dimension predicted vector, and the n is tieed up into pre- direction finding
Amount regards as neural predicted vector.Wherein, n is positive integer, and soft-max classifier is regression model in more classification problems
It promotes and applies.
The prediction knot of cluster result and default convolutional neural networks based on context semanteme will be combined in the present embodiment
Fruit carries out personality prediction, specifically, the ginseng for being directed to context semanteme will be carried out based on semantic weight and K-means clustering algorithm
It examines, meanwhile, the convolutional neural networks of Text-CNN will be used to carry out personality prediction.
It is to be based on the present invention is based on the flow diagram of semantic personality prediction technique 3rd embodiment referring to Fig. 4, Fig. 4
Above-mentioned first embodiment shown in Fig. 2 proposes that the present invention is based on the 3rd embodiments of semantic personality prediction technique.
In 3rd embodiment, the step S70, comprising:
Step S701: according to the vector to be entered by preset classifier carry out user's personality prediction, with obtain with
The corresponding personality probability of user's personality.
It is understood that the prediction of cluster result and default convolutional neural networks based on context semanteme can be combined
As a result integrated prediction is carried out, to realize comprehensive a variety of considerations to complete the predicted operation of personality prediction.
In the concrete realization, it if using the foundation that five-factor model personality theory is predicted as personality, can obtain respectively in this 5 kinds of personalities
Personality probability, personality probability is for a possibility that whether user is the personality type to be recorded.In view of five kinds of personalities are respectively
Open personality, sense of responsibility personality, extroversion personality, pleasant personality and neurotic personality, may calculated open people
The corresponding personality probability of lattice is 0.6, the corresponding personality probability of calculated sense of responsibility personality is 0.3, calculated extroversion people
The corresponding personality probability of lattice is 0.7, the corresponding personality probability of calculated pleasant personality is 0.3 and calculated neurotic people
The corresponding personality probability of lattice is 0.4.
Step S702: when the personality probability is more than or equal to predetermined probabilities threshold value, it will be greater than the predetermined probabilities threshold value
The corresponding user's personality of personality probability as personality prediction result.
It should be understood that it is 0.5 that predetermined probabilities threshold value, which can be preset, it will be apparent that, open personality and extroversion
The corresponding personality probability of personality is greater than 0.5, then the personality prediction result of user is open personality and extroversion personality.
Further, the prediction for carrying out user's personality by presetting classifier according to the vector to be entered, to obtain
Before obtaining personality prediction result, the semantic-based personality prediction technique is further comprising the steps of:
It obtains to training vector；
Based on Adaboost algorithm it is determining with described to the corresponding initial weight of training vector；
It is trained to described to training vector based on the initial weight, to obtain the first Weak Classifier；
Determine the first error rate corresponding with first Weak Classifier；
When first error rate is less than or equal to default error rate threshold, the Weak Classifier of acquisition is combined, with
Combination obtains strong classifier, and the strong classifier is regarded as default classifier.
It is understood that in order to obtain for personality prediction default classifier, can be used Adaboost algorithm come into
Integrated, the accuracy predicted with raising of row classifier.
It should be understood that vector to be entered can be used directly to determine corresponding strong classifier, so, to training vector
It can be identical as vector to be entered；Can also be used it is different determine corresponding strong classifier to training vector, getting strong point
It after class device, is directly handled using the strong classifier to training vector, to eliminate the training process of strong classifier.
In the concrete realization, it is obtaining after training vector, Adaboost (adaptive boosting) can first be used to calculate
Method is to distribute a numerical value identical initial weight to each of training vector sample.Wherein, Adaboost algorithm is one
Kind iterative algorithm, specifically, Adaboost.MH algorithm can train different Weak Classifiers for the same training set, so
Afterwards, these weak classifier sets are got up and obtains a strong classifier.
It should be understood that training representated by training vector can be treated based on initial weight after obtaining initial weight
Collection is trained, and by available first Weak Classifier of training, and calculates first error rates of weak classifiers, error rate is used for
The accuracy that assessment Weak Classifier is differentiated.If the error rate is less than default error rate threshold, it is believed that the Weak Classifier
Meet and differentiates requirement.So integrated combination can be carried out to acquired all Weak Classifiers, the Weak Classifier after combination is as strong
Classifier.
It should be noted that if the first Weak Classifier has met accuracy requirement, then the Weak Classifier obtained is first weak
Classifier.
Further, described based on semanteme after the determination the first error rate corresponding with first Weak Classifier
Personality prediction technique it is further comprising the steps of:
When first error rate is greater than the default error rate threshold, the initial weight is adjusted, to obtain
Obtain the second weight；
It is trained to described to training vector based on second weight, to obtain the second Weak Classifier；
Determine the second error rate corresponding with second Weak Classifier；
When second error rate is less than or equal to the default error rate threshold, the Weak Classifier of acquisition is subjected to group
It closes, strong classifier is obtained with combination, and the strong classifier is regarded as into default classifier.
It is understood that the second Weak Classifier will be generated if the first Weak Classifier is unsatisfactory for accuracy requirement.Specifically
For, initial weight will be adjusted, for example, if there is misjudgement, there is also correct to multiple samples in training vector
Determine, then can will be to determine that the corresponding initial weight of correct sample carries out multiplying an operation in training vector, will be to training vector
The corresponding initial weight of the sample of middle decision error carries out multiplying negative one operation, to obtain the weight after operation, and will be after operation
Weight regards as the second weight.It, can be weak based on the acquisition second of the second weight is stated after obtaining new weight distribution and being the second weight
Classifier.
It should be understood that if the second Weak Classifier meets accuracy requirement, can by the first Weak Classifier of acquisition with
Second Weak Classifier is combined to obtain strong classifier.It certainly, can be by initial weight and the second weight when being combined
Operation is normalized.
Certainly, if the second Weak Classifier is still unsatisfactory for accuracy requirement i.e. the second error rate and is greater than default error rate threshold,
Second weight can be then adjusted, to obtain third weight；Based on the third weight to it is described to training vector into
Row training, to obtain third Weak Classifier；Determine third error rate corresponding with the third Weak Classifier；It is wrong in the third
When accidentally rate is less than or equal to the default error rate threshold, the Weak Classifier of acquisition is combined, strong classifier is obtained with combination,
And the strong classifier is regarded as into default classifier.And so on, it is recycled.
The prediction knot of cluster result and default convolutional neural networks based on context semanteme is combined in the present embodiment
Fruit improves Clustering Effect, improves the accuracy of personality prediction.Moreover, it is contemplated that the people in current most of social networks
What lattice prediction used is all single classifier, and integrated classifier is rarely employed, and the prediction result accuracy obtained in this way is not high enough.
So will be predicted using integrated classifier in the present embodiment, the accuracy of personality prediction is further improved, solves personality
The lower technical problem of predictablity rate.
In addition, the embodiment of the present invention also proposes a kind of storage medium, semantic-based people is stored on the storage medium
Lattice Prediction program realizes following operation when the semantic-based personality Prediction program is executed by processor:
To in text set to be predicted text to be predicted carry out word segmentation processing, with obtain in the text to be predicted respectively to
Operate text feature word；
Calculate semantic weight corresponding with the text feature word to be operated；
The text vector being made of the semantic weight is clustered, to obtain Semantic Clustering vector；
Distributed expression processing is carried out to the text to be predicted, to obtain term vector；
The term vector is trained based on default convolutional neural networks, to obtain neural predicted vector；
Institute's semantic cluster vector and the neural predicted vector are spliced, to obtain vector to be entered；
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain personality prediction knot
Fruit.
Further, following operation is also realized when the semantic-based personality Prediction program is executed by processor:
Default frequency of occurrence of the co-occurrence Feature Words in the text set to be predicted is counted, the default co-occurrence Feature Words are
The two text feature words to be operated occurred in a text to be predicted；
According to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence.
Further, following operation is also realized when the semantic-based personality Prediction program is executed by processor:
According to the semantic weight construction feature word matrix；
Singular value decomposition is carried out to the Feature Words matrix, to obtain semantic matrix；
Text vector in the semantic matrix is clustered, to obtain Semantic Clustering vector.
Further, following operation is also realized when the semantic-based personality Prediction program is executed by processor:
Convolution sum pond is carried out to the term vector based on default convolutional neural networks, to obtain feature vector；
Default word frequency vector is spliced with described eigenvector, to obtain splicing vector；
The splicing vector is trained, to obtain prediction probability value, and prediction probability value composition nerve is pre-
Direction finding amount.
Further, following operation is also realized when the semantic-based personality Prediction program is executed by processor:
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain and the user people
The corresponding personality probability of lattice；
When the personality probability is more than or equal to predetermined probabilities threshold value, the personality probability of the predetermined probabilities threshold value will be greater than
Corresponding user's personality is as personality prediction result.
Further, following operation is also realized when the semantic-based personality Prediction program is executed by processor:
It obtains to training vector；
Based on Adaboost algorithm it is determining with described to the corresponding initial weight of training vector；
It is trained to described to training vector based on the initial weight, to obtain the first Weak Classifier；
Determine the first error rate corresponding with first Weak Classifier；
When first error rate is less than or equal to default error rate threshold, the Weak Classifier of acquisition is combined, with
Combination obtains strong classifier, and the strong classifier is regarded as default classifier.
Further, following operation is also realized when the semantic-based personality Prediction program is executed by processor:
When first error rate is greater than the default error rate threshold, the initial weight is adjusted, to obtain
Obtain the second weight；
It is trained to described to training vector based on second weight, to obtain the second Weak Classifier；
Determine the second error rate corresponding with second Weak Classifier；
When second error rate is less than or equal to the default error rate threshold, the Weak Classifier of acquisition is subjected to group
It closes, strong classifier is obtained with combination, and the strong classifier is regarded as into default classifier.
The text feature word respectively to be operated in text to be predicted is first obtained in the present embodiment, is calculated and text feature to be operated
The corresponding semantic weight of word clusters the text vector being made of semantic weight to obtain Semantic Clustering vector；It treats again
Predict text carry out it is distributed indicate processing to obtain term vector, based on default convolutional neural networks to term vector be trained with
Obtain neural predicted vector；Semantic Clustering vector and neural predicted vector are spliced, to obtain vector to be entered, according to
Input vector carries out the prediction of user's personality by presetting classifier.It is apparent that having combined in the present embodiment based on context language
The cluster result of justice and the prediction result of default convolutional neural networks, improve Clustering Effect, improve the standard of personality prediction
True property solves the lower technical problem of personality predictablity rate.
In addition, the embodiment of the present invention also proposes a kind of semantic-based personality prediction meanss, described to be based on language referring to Fig. 5
Justice personality prediction meanss include:
Word segmentation processing module 10, for carrying out word segmentation processing to the text to be predicted in text set to be predicted, to obtain
State the text feature word respectively to be operated in text to be predicted.
It is understood that can first be obtained from social networks big to realize that the personality under social networks scene is predicted
The text to be predicted of amount, text set to be predicted are denoted as the set of a large amount of text to be predicted.Text to be predicted can be user A in society
The text information generated in network is handed over, the personality prediction of user A can be carried out based on the text to be predicted.
In the concrete realization, first the text to be predicted can be pre-processed, by text dividing to be predicted at single one by one
Only word.
Semantic weight computing module 20, for calculating semantic weight corresponding with the text feature word to be operated.
It should be understood that semantic weight corresponding with the text feature word to be operated being syncopated as, semantic weight can be calculated
For characterizing the correlation degree between context semanteme.
Cluster module 30, for being clustered to the text vector being made of the semantic weight, to obtain Semantic Clustering
Vector.
It is understood that the text vector being made of semantic weight can be clustered based on K-means clustering algorithm,
To obtain Semantic Clustering vector.Wherein, it is contemplated that each text feature word is there is corresponding semantic weight, and text feature
Word is combined into short, the semantic weight of each text feature word of a word can be constituted one-dimensional text vector.In addition,
K-means clustering algorithm belongs to a kind of typical hard clustering algorithm.
Distributed processing modules 40, for carrying out distributed expression processing to the text to be predicted, to obtain term vector.
It is understood that step S10-S30 is by handling text to be predicted, the Semantic Clustering handled to
Amount can characterize the correlation degree between context semanteme.But other than the cluster for context semanteme, it can also pass through step
Rapid S40-50 carries out the prediction of convolutional neural networks, by with reference to the relevance and convolutional Neural between context semanteme
The prediction result of network can preferably improve the accuracy of prediction result.
In the concrete realization, it in order to carry out personality prediction based on convolutional neural networks, can first text to be predicted be distributed
Formula expression processing, textual form to be predicted, which is turned to vector, to be indicated, to obtain term vector.
Neural network module 50, for being trained based on default convolutional neural networks to the term vector, to obtain mind
Predicted vector.
It should be understood that term vector can be inputted to the input layer of default convolutional neural networks, led to after obtaining term vector
Training is crossed to realize the neural predicted vector of final acquisition.
Vector splicing module 60, for splicing to institute's semantic cluster vector and the neural predicted vector, to obtain
Obtain vector to be entered.
It is understood that the Semantic Clustering vector that context Semantic Clustering obtains may tie up output vector for m, volume is preset
The neural predicted vector that product neural network prediction obtains may tie up output vector for n, and m and n are positive integer, can combination semantic cluster
Vector and neural predicted vector carry out integrated prediction.
Personality prediction module 70, for carrying out the pre- of user's personality by presetting classifier according to the vector to be entered
It surveys, to obtain personality prediction result.
It should be understood that the vector that will be tieed up by the vector to be entered that concatenation obtains for m+n, and the vector is returned
Input vector after one change as integrated prediction.
In the concrete realization, if using the foundation that five-factor model personality theory is predicted as personality, can input into default classifier should
Vector to be entered carries out the prediction of the personality based on five-factor model personality by default classifier, to export personality prediction result.Wherein,
Personality prediction result may be the personality or multinomial personality in five-factor model personality theory.
The text feature word respectively to be operated in text to be predicted is first obtained in the present embodiment, is calculated and text feature to be operated
The corresponding semantic weight of word clusters the text vector being made of semantic weight to obtain Semantic Clustering vector；It treats again
Predict text carry out it is distributed indicate processing to obtain term vector, based on default convolutional neural networks to term vector be trained with
Obtain neural predicted vector；Semantic Clustering vector and neural predicted vector are spliced, to obtain vector to be entered, according to
Input vector carries out the prediction of user's personality by presetting classifier.It is apparent that having combined in the present invention semantic based on context
Cluster result and default convolutional neural networks prediction result, improve Clustering Effect, improve personality prediction it is accurate
Property, solve the lower technical problem of personality predictablity rate.
In one embodiment, the semantic weight computing module 20, be also used to count default co-occurrence Feature Words it is described to
Predict the frequency of occurrence in text set, the default co-occurrence Feature Words be occur in a text to be predicted two to
Operate text feature word；According to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence.
In one embodiment, the cluster module 30 is also used to according to the semantic weight construction feature word matrix；To institute
It states Feature Words matrix and carries out singular value decomposition, to obtain semantic matrix；Text vector in the semantic matrix is clustered,
To obtain Semantic Clustering vector.
In one embodiment, the neural network module 50, be also used to based on default convolutional neural networks to institute's predicate to
Amount carries out convolution sum pond, to obtain feature vector；Default word frequency vector is spliced with described eigenvector, to be spelled
Connect vector；The splicing vector is trained, to obtain prediction probability value, and the prediction probability value is formed into nerve prediction
Vector.
In one embodiment, the personality prediction module 70 is also used to pass through default classification according to the vector to be entered
Device carries out the prediction of user's personality, to obtain personality probability corresponding with user's personality；Be greater than in the personality probability etc.
When predetermined probabilities threshold value, the corresponding user's personality of personality probability that will be greater than the predetermined probabilities threshold value is predicted to tie as personality
Fruit.
In one embodiment, the semantic-based personality prediction meanss further include:
First integrated classifier generation module, for obtaining to training vector；Based on Adaboost algorithm it is determining with it is described
To the corresponding initial weight of training vector；It is trained to described to training vector based on the initial weight, to obtain first
Weak Classifier；Determine the first error rate corresponding with first Weak Classifier；It is less than or equal in first error rate default
When error rate threshold, the Weak Classifier of acquisition is combined, strong classifier is obtained with combination, and the strong classifier is assert
To preset classifier.
In one embodiment, the semantic-based personality prediction meanss further include:
Second integrated classifier generation module is used for when first error rate is greater than the default error rate threshold,
The initial weight is adjusted, to obtain the second weight；It is instructed to described to training vector based on second weight
Practice, to obtain the second Weak Classifier；Determine the second error rate corresponding with second Weak Classifier；In second error rate
When less than or equal to the default error rate threshold, the Weak Classifier of acquisition is combined, strong classifier is obtained with combination, and will
The strong classifier regards as default classifier.
The other embodiments or specific implementation of semantic-based personality prediction meanss of the present invention can refer to above-mentioned
Each method embodiment, details are not described herein again.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the system that include a series of elements not only include those elements, and
And further include other elements that are not explicitly listed, or further include for this process, method, article or system institute it is intrinsic
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including being somebody's turn to do
There is also other identical elements in the process, method of element, article or system.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.If listing equipment for drying
Unit claim in, several in these devices, which can be, to be embodied by the same item of hardware.Word first,
Second and the use of third etc. do not indicate any sequence, can be title by these word explanations.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in a storage medium
In (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, computer, clothes
Business device, air conditioner or the network equipment etc.) execute method described in each embodiment of the present invention.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of semantic-based personality prediction technique, which is characterized in that the semantic-based personality prediction technique include with
Lower step:
To in text set to be predicted text to be predicted carry out word segmentation processing, with obtain in the text to be predicted respectively wait operate
Text feature word；
Calculate semantic weight corresponding with the text feature word to be operated；
The text vector being made of the semantic weight is clustered, to obtain Semantic Clustering vector；
Distributed expression processing is carried out to the text to be predicted, to obtain term vector；
The term vector is trained based on default convolutional neural networks, to obtain neural predicted vector；
Institute's semantic cluster vector and the neural predicted vector are spliced, to obtain vector to be entered；
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain personality prediction result.
2. semantic-based personality prediction technique as described in claim 1, which is characterized in that the calculating is with described wait operate
The corresponding semantic weight of text feature word, comprising:
Default frequency of occurrence of the co-occurrence Feature Words in the text set to be predicted is counted, the default co-occurrence Feature Words are one
The two text feature words to be operated occurred in a text to be predicted；
According to the determining semantic weight corresponding with the default co-occurrence Feature Words of the frequency of occurrence.
3. semantic-based personality prediction technique as described in claim 1, which is characterized in that described to by the semantic weight
The text vector of composition is clustered, to obtain Semantic Clustering vector, comprising:
According to the semantic weight construction feature word matrix；
Singular value decomposition is carried out to the Feature Words matrix, to obtain semantic matrix；
Text vector in the semantic matrix is clustered, to obtain Semantic Clustering vector.
4. semantic-based personality prediction technique as claimed any one in claims 1 to 3, which is characterized in that described to be based on
Default convolutional neural networks are trained the term vector, to obtain neural predicted vector, comprising:
Convolution sum pond is carried out to the term vector based on default convolutional neural networks, to obtain feature vector；
Default word frequency vector is spliced with described eigenvector, to obtain splicing vector；
The splicing vector is trained, to obtain prediction probability value, and the prediction probability value is formed into the pre- direction finding of nerve
Amount.
5. semantic-based personality prediction technique as claimed any one in claims 1 to 3, which is characterized in that the basis
The vector to be entered carries out the prediction of user's personality by presetting classifier, to obtain personality prediction result, comprising:
The prediction of user's personality is carried out, by presetting classifier according to the vector to be entered to obtain and user's personality pair
The personality probability answered；
When the personality probability is more than or equal to predetermined probabilities threshold value, the personality probability that will be greater than the predetermined probabilities threshold value is corresponding
User's personality as personality prediction result.
6. semantic-based personality prediction technique as claimed any one in claims 1 to 3, which is characterized in that the basis
The vector to be entered carries out the prediction of user's personality, before obtaining personality prediction result, the base by presetting classifier
It is further comprising the steps of in semantic personality prediction technique:
It obtains to training vector；
Based on Adaboost algorithm it is determining with described to the corresponding initial weight of training vector；
It is trained to described to training vector based on the initial weight, to obtain the first Weak Classifier；
Determine the first error rate corresponding with first Weak Classifier；
When first error rate is less than or equal to default error rate threshold, the Weak Classifier of acquisition is combined, with combination
Strong classifier is obtained, and the strong classifier is regarded as into default classifier.
7. semantic-based personality prediction technique as claimed in claim 6, which is characterized in that the determination and described first weak
After corresponding first error rate of classifier, the semantic-based personality prediction technique is further comprising the steps of:
When first error rate is greater than the default error rate threshold, the initial weight is adjusted, to obtain the
Two weights；
It is trained to described to training vector based on second weight, to obtain the second Weak Classifier；
Determine the second error rate corresponding with second Weak Classifier；
When second error rate is less than or equal to the default error rate threshold, the Weak Classifier of acquisition is combined, with
Combination obtains strong classifier, and the strong classifier is regarded as default classifier.
8. a kind of user equipment, which is characterized in that the user equipment includes: memory, processor and is stored in the storage
Semantic-based personality Prediction program, the semantic-based personality Prediction program quilt can be run on device and on the processor
The step of semantic-based personality prediction technique as described in any one of claims 1 to 7 is realized when the processor executes.
9. a kind of storage medium, which is characterized in that semantic-based personality Prediction program is stored on the storage medium, it is described
When semantic-based personality Prediction program is executed by processor realize as described in any one of claims 1 to 7 based on semanteme
Personality prediction technique the step of.
10. a kind of semantic-based personality prediction meanss, which is characterized in that the semantic-based personality prediction meanss include:
Word segmentation processing module, it is described to pre- to obtain for carrying out word segmentation processing to the text to be predicted in text set to be predicted
Survey the text feature word respectively to be operated in text；
Semantic weight computing module, for calculating semantic weight corresponding with the text feature word to be operated；
Cluster module, for being clustered to the text vector being made of the semantic weight, to obtain Semantic Clustering vector；
Distributed processing modules, for carrying out distributed expression processing to the text to be predicted, to obtain term vector；
Neural network module, for being trained based on default convolutional neural networks to the term vector, to obtain nerve prediction
Vector；
Vector splicing module, for splicing to institute's semantic cluster vector and the neural predicted vector, to obtain to defeated
Incoming vector；
Personality prediction module, for carrying out the prediction of user's personality by presetting classifier according to the vector to be entered, to obtain
Obtain personality prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910042095.6A CN109829154B (en) | 2019-01-16 | 2019-01-16 | Personality prediction method based on semantics, user equipment, storage medium and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910042095.6A CN109829154B (en) | 2019-01-16 | 2019-01-16 | Personality prediction method based on semantics, user equipment, storage medium and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109829154A true CN109829154A (en) | 2019-05-31 |
| CN109829154B CN109829154B (en) | 2023-04-28 |
Family
ID=66861711
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910042095.6A Active CN109829154B (en) | 2019-01-16 | 2019-01-16 | Personality prediction method based on semantics, user equipment, storage medium and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109829154B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110275953A (en) * | 2019-06-21 | 2019-09-24 | 四川大学 | Personality classification method and device |
| CN111797194A (en) * | 2020-05-20 | 2020-10-20 | 北京三快在线科技有限公司 | Text risk detection method and device, electronic equipment and storage medium |
| CN112487184A (en) * | 2020-11-26 | 2021-03-12 | 北京智源人工智能研究院 | User character judging method and device, memory and electronic equipment |
| CN112650861A (en) * | 2020-12-29 | 2021-04-13 | 中山大学 | Personality prediction method, system and device based on task layering |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899608A (en) * | 2015-06-19 | 2015-09-09 | 西安交通大学 | Weight fusion relevant vector machine model for prediction of residual life of antifriction bearing |
| WO2018060993A1 (en) * | 2016-09-27 | 2018-04-05 | Faception Ltd. | Method and system for personality-weighted emotion analysis |
| US20180203848A1 (en) * | 2017-01-17 | 2018-07-19 | Xerox Corporation | Author personality trait recognition from short texts with a deep compositional learning approach |
| CN108399575A (en) * | 2018-01-24 | 2018-08-14 | 大连理工大学 | A kind of five-factor model personality prediction technique based on social media text |
-
2019
- 2019-01-16 CN CN201910042095.6A patent/CN109829154B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899608A (en) * | 2015-06-19 | 2015-09-09 | 西安交通大学 | Weight fusion relevant vector machine model for prediction of residual life of antifriction bearing |
| WO2018060993A1 (en) * | 2016-09-27 | 2018-04-05 | Faception Ltd. | Method and system for personality-weighted emotion analysis |
| US20180203848A1 (en) * | 2017-01-17 | 2018-07-19 | Xerox Corporation | Author personality trait recognition from short texts with a deep compositional learning approach |
| CN108399575A (en) * | 2018-01-24 | 2018-08-14 | 大连理工大学 | A kind of five-factor model personality prediction technique based on social media text |
Non-Patent Citations (1)
| Title |
|---|
| WEI H等: "Beyond the words: Predicting user personality from heterogeneous information", 《PROCEEDINGS OF THE 10TH ACM INTERNATIONAL》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110275953A (en) * | 2019-06-21 | 2019-09-24 | 四川大学 | Personality classification method and device |
| CN110275953B (en) * | 2019-06-21 | 2021-11-30 | 四川大学 | Personality classification method and apparatus |
| CN111797194A (en) * | 2020-05-20 | 2020-10-20 | 北京三快在线科技有限公司 | Text risk detection method and device, electronic equipment and storage medium |
| CN111797194B (en) * | 2020-05-20 | 2024-04-02 | 北京三快在线科技有限公司 | Text risk detection method and device, electronic equipment and storage medium |
| CN112487184A (en) * | 2020-11-26 | 2021-03-12 | 北京智源人工智能研究院 | User character judging method and device, memory and electronic equipment |
| CN112650861A (en) * | 2020-12-29 | 2021-04-13 | 中山大学 | Personality prediction method, system and device based on task layering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109829154B (en) | 2023-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | A recommendation model based on deep neural network | |
| CN108182279B (en) | Object classification method, device and computer equipment based on text feature | |
| Bi et al. | A big data clustering algorithm for mitigating the risk of customer churn | |
| CN109829154A (en) | Semantic-based personality prediction technique, user equipment, storage medium and device | |
| Kunegis et al. | The link prediction problem in bipartite networks | |
| CN110032632A (en) | Intelligent customer service answering method, device and storage medium based on text similarity | |
| CN112632385A (en) | Course recommendation method and device, computer equipment and medium | |
| CN107862022B (en) | Culture resource recommendation system | |
| US20160357845A1 (en) | Method and Apparatus for Classifying Object Based on Social Networking Service, and Storage Medium | |
| US20220172260A1 (en) | Method, apparatus, storage medium, and device for generating user profile | |
| Zhang et al. | Unifying paragraph embeddings and neural collaborative filtering for hybrid recommendation | |
| CN112307332A (en) | Collaborative filtering recommendation method and system based on user portrait clustering and storage medium | |
| CN114077661A (en) | Information processing apparatus, information processing method, and computer readable medium | |
| CN112215629B (en) | Multi-target advertisement generating system and method based on construction countermeasure sample | |
| CN108595411A (en) | More text snippet acquisition methods in a kind of same subject text set | |
| Tu et al. | Bidirectional sensing of user preferences and application changes for dynamic mobile app recommendations | |
| CN111061876A (en) | Event public opinion data analysis method and device | |
| Najar et al. | Exact fisher information of generalized dirichlet multinomial distribution for count data modeling | |
| CN111737576B (en) | Application function personalized recommendation method and device | |
| CN113641821A (en) | Value orientation identification method and system for opinion leaders in social network | |
| Deekshan et al. | Detection and summarization of honest reviews using text mining | |
| Saha et al. | A large scale study of SVM based methods for abstract screening in systematic reviews | |
| CN110085292A (en) | Drug recommended method, device and computer readable storage medium | |
| CN106446696A (en) | Information processing method and electronic device | |
| CN114693409A (en) | Product matching method, device, computer equipment, storage medium and program product |
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 |