CN107169035A - A kind of file classification method for mixing shot and long term memory network and convolutional neural networks - Google Patents

A kind of file classification method for mixing shot and long term memory network and convolutional neural networks Download PDF

Info

Publication number
CN107169035A
CN107169035A CN201710257132.6A CN201710257132A CN107169035A CN 107169035 A CN107169035 A CN 107169035A CN 201710257132 A CN201710257132 A CN 201710257132A CN 107169035 A CN107169035 A CN 107169035A
Authority
CN
China
Prior art keywords
sentence
layer
convolutional neural
long term
memory network
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201710257132.6A
Other languages
Chinese (zh)
Other versions
CN107169035B (en
Inventor
苏锦钿
霍振朗
欧阳志凡
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN201710257132.6A priority Critical patent/CN107169035B/en
Publication of CN107169035A publication Critical patent/CN107169035A/en
Application granted granted Critical
Publication of CN107169035B publication Critical patent/CN107169035B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Abstract

The invention discloses a kind of file classification method for mixing shot and long term memory network and convolutional neural networks, by fully combining two-way shot and long term memory network in the advantage of advantage and convolutional neural networks in terms of learning text local feature in terms of the contextual information of learning text, after the contextual information for learning word using two-way shot and long term memory network, further learn to extract the local feature of the term vector of contextual information by convolutional neural networks again, then two-way shot and long term memory network is recycled to learn the context of these local features, form the output of fixed dimension, classification output is carried out finally by a multilayer perceptron.The accuracy rate of category of model can be further improved, and with preferable versatility, good effect is all achieved on multiple corpus of test.

Description

A kind of file classification method for mixing shot and long term memory network and convolutional neural networks
Technical field
The present invention relates to natural language processing field, and in particular to one kind mixing shot and long term memory network and convolutional Neural net The file classification method of network.
Background technology
Text automatic classification based on machine learning is a research most popular in natural language processing field in recent years Direction, it is numerous in information retrieval, search engine, automatic question answering, ecommerce, digital library, automatic abstract, news portal etc. Field has obtained extensive application.So-called text automatic classification refers on the premise of given taxonomic hierarchies, sharp The process of text categories is automatically determined after being analyzed with the mode of machine learning come the content to text.1990s with Before, text automatic classification mainly by the way of KBE, i.e., is classified, it has the disadvantage into by hand by professional This height, waste time and energy.Since the nineties, many researchers start various statistical methods and machine learning method being applied to certainly Dynamic text classification, such as support vector machines, AdaBoost algorithms, NB Algorithm, KNN algorithms and Logistic are returned Return.In recent years, with the fast development of deep learning and various neural network models, the text classification side based on deep learning Method causes the close attention and research of academia and industrial quarters, some typical neural network models, such as recurrent neural network (using shot and long term memory network LSTM and GRU as main representative) and convolutional neural networks CNN are widely used in text In classification, and achieve good effect.Existing research and application are proved recurrent neural network and are suitable for learning sentence Long-term dependence between middle linguistic unit, convolutional neural networks are suitable for learning the local feature of sentence, but current grind Study carefully and be not adequately bonded recurrent neural network and the respective advantage of convolutional neural networks, be also not bound with language in consideration sentence The contextual information of unit.
The content of the invention
The purpose of the present invention is that there is provided one kind mixing shot and long term memory network and volume for above-mentioned the deficiencies in the prior art The file classification method of product neutral net, using the information above and context information of word in two-way LSTM learning texts sentence, connects Learning outcome and local feature is further extracted by CNN, then recycle one two-way LSTM layers to learn local feature Between relation, finally learning outcome is classified and exported by a multilayer perceptron.
The purpose of the present invention can be achieved through the following technical solutions：
A kind of file classification method for mixing shot and long term memory network and convolutional neural networks, methods described includes following step Suddenly：
Step 1, the sentence in text is pre-processed, the distribution of lengths of sentence and square in combined training corpus Unified sentence length is formed after difference, the length threshold for determining sentence, input text is obtained using the good term vector table of pre-training In the vectorization of each word represent, form continuous and dense real number vector matrix；
Step 2, the sentence term vector for input, respectively by positive each word of LSTM e-learnings above The context information of information and reverse each word of LSTM e-learnings, and the result of study is subjected to series connection merging, so that Sentence term vector comprising semantic information is represented to be converted into while including the expression of semantic and contextual information；
Step 3, the word exported respectively to two-way LSTM networks using multiple different in width, the nuclear matrix comprising different weights Vector matrix carries out two-dimensional convolution computing, extracts local convolution feature, and generate the local convolution eigenmatrix of multilayer；
Step 4, using one-dimensional maximum pond algorithm down-sampling is carried out to the local convolution eigenmatrix of multilayer, obtain sentence Multilayer global characteristics matrix, and by result carry out series connection merging；
Step 5, using the LSTM networks of two opposite directions learn the long-term dependence between sentence local feature respectively, And exported last learning outcome；
Step 6, the output result of step 5 is first passed through to a full connection hidden layer, it is then right by one softmax layers again The classification of sentence is predicted.
Further, a kind of shot and long term memory network and the file classification method of convolutional neural networks of mixing is one Completed in individual multilayer neural network, the step 1 is completed in first layer input layer, and step 2 is two-way LSTM layers in the second layer Middle to complete, step 3 completion in CNN layer of third layer, step 4 is completed in the 4th layer of pond layer, and step 5 is two-way in layer 5 Completed in LSTM layers, step 6 is completed in layer 6 output layer.
Further, two-way LSTM layers of the second layer is used for the context letter for learning to be originally inputted the word of each in sentence Breath, and export after the learning outcome of each word is connected, two-way LSTM layer of the layer 5 learns sentence spy after convolution Contextual information between levying, and only export the learning outcome of final step.
Further, it is described to include punctuation mark filtering, abbreviation polishing to sentence progress pretreatment, delete sky in step 1 Lattice, participle and forbidden character are carried out to sentence filter.
Further, the step 3 is local feature learning process, passes through the two-dimensional convolution window of multiple different word step-lengths The term vector comprising contextual information is learnt with convolution kernel, so as to obtain varigrained phrase information.
Further, the step 4 is sampling and reduction process, and multilayer is locally rolled up by one-dimensional maximum pond algorithm Product eigenmatrix carries out down-sampling, obtains most important characteristic value in the window of the pond of each in sentence, and be used as local window In character representation.
Further, the step 5 learns for the context of local feature, by between two-way LSTM study local features Contextual information, and the learning outcome of last term vector is exported, while forming the one-dimensional output of fixed dimension.
Further, the step 6 is exported for classification, and classification judgement is carried out by a multilayer perceptron connected entirely, And final output is obtained according to the probability distribution situation on specified taxonomic hierarchies.
Further, the step 6 is completed in the multilayer perceptron of two layers, including a full connection hidden layer and one Individual softmax layers, the output result of step 6 is the prediction classification of correspondence text.
The present invention compared with prior art, has the following advantages that and beneficial effect：
The present invention is being learned by fully combining advantages and CNN of the two-way LSTM in terms of the contextual information of learning text The advantage in terms of text local feature is practised, a kind of mixing LSTM and CNN file classification method is proposed, by using two-way LSTM Learn word contextual information after, then by CNN further learn extract contextual information term vector local feature, then Recycle two-way LSTM to learn the context of these local features, the output of fixed dimension is formed, finally by a multilayer sense Know that device carries out classification output.The accuracy rate of category of model can be further improved, and with preferable versatility, in many of test Good effect is all achieved on individual corpus.
Brief description of the drawings
Fig. 1 is the general frame figure of multilayer neural network model of the embodiment of the present invention.
Embodiment
With reference to embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited In this.
Embodiment：
A kind of file classification method for mixing shot and long term memory network and convolutional neural networks is present embodiments provided, it is described Method comprises the following steps：
Step 1, sentence in text is pre-processed, including punctuation mark filtering, abbreviation polishing, delete space, distich Son carries out the distribution of lengths and mean square deviation of sentence in participle and forbidden character filtering, combined training corpus, determines the length of sentence Unified sentence length is formed after degree threshold value, the vector of each word in input text is obtained using the good term vector table of pre-training Change and represent, form continuous and dense real number vector matrix；
Step 2, the sentence term vector for input, respectively by positive each word of LSTM e-learnings above The context information of information and reverse each word of LSTM e-learnings, and the result of study is subjected to series connection merging, so that Sentence term vector comprising semantic information is represented to be converted into while including the expression of semantic and contextual information；
Step 3, the word exported respectively to two-way LSTM networks using multiple different in width, the nuclear matrix comprising different weights Vector matrix carries out two-dimensional convolution computing, extracts local convolution feature, and generate the local convolution eigenmatrix of multilayer；
Step 4, using one-dimensional maximum pond algorithm down-sampling is carried out to the local convolution eigenmatrix of multilayer, obtain sentence Multilayer global characteristics matrix, and by result carry out series connection merging；
Step 5, using the LSTM networks of two opposite directions learn the long-term dependence between sentence local feature respectively, And exported last learning outcome；
Step 6, the output result of step 5 is first passed through to a full connection hidden layer, it is then right by one softmax layers again The classification of sentence is predicted.
A kind of shot and long term memory network and the file classification method of convolutional neural networks of mixing described above is more than one Completed in layer neutral net, the Organization Chart of multilayer neural network is as shown in figure 1, the step 1 is complete in first layer input layer Into；Step 2 is completed in the second layer is two-way LSTM layers, wherein, two-way LSTM output dimension is 256 dimensions；Step 3 is in third layer Completed in CNN layers, wherein, the convolution word step-length in CNN layers is respectively 2,3,4, and output dimension is 128 dimensions；Step 4 is at the 4th layer Completed in the layer of pond, the word step-length of pond window is respectively 2,3,4, and use one-dimensional maximum pond；Step 5 is in layer 5 Completed in two-way LSTM layers, wherein two-way LSTM layers output dimension is 128 dimensions, and only export the study knot of last word Really；Step 6 is completed in layer 6 output layer, and the output layer is the multilayer perceptron of two layers, including a full connection Hidden layer and one softmax layers, the full connection hidden layer is 128 dimensions, and dropout values are 0.5, and the output result of step 6 is pair Answer the prediction classification of text.Loss function is defined using polynary cross entropy during model training, and combines RMSProp optimizations Device.
Wherein, two-way LSTM layers of the second layer is used for the contextual information for learning to be originally inputted the word of each in sentence, and And exported after the learning outcome of each word is connected, after the two-way LSTM layers of study convolution of layer 5 between sentence characteristics Contextual information, and only export the learning outcome of final step.
Wherein, the step 3 is local feature learning process, passes through the two-dimensional convolution window and volume of multiple different word step-lengths Term vector of the product verification comprising contextual information is learnt, so as to obtain varigrained phrase information, the step 4 is to adopt Sample and reduction process, carry out down-sampling to the local convolution eigenmatrix of multilayer by one-dimensional maximum pond algorithm, obtain sentence In most important characteristic value in each pond window, and as the character representation in local window, the step 5 is local special The context study levied, learns the contextual information between local feature, and export last term vector by two-way LSTM Learning outcome, while formed fixed dimension one-dimensional output, the step 6 for classification export, by one connect entirely it is many Layer perceptron carries out classification judgement, and obtains final output according to the probability distribution situation on specified taxonomic hierarchies.
It is described above, it is only patent preferred embodiment of the present invention, but the protection domain of patent of the present invention is not limited to This, any one skilled in the art is in the scope disclosed in patent of the present invention, according to the skill of patent of the present invention Art scheme and its patent of invention design are subject to equivalent substitution or change, belong to the protection domain of patent of the present invention.

Claims (9)

1. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks, it is characterised in that methods described Comprise the following steps：
Step 1, the sentence in text is pre-processed, the distribution of lengths and mean square deviation of sentence in combined training corpus, really Unified sentence length is formed after the length threshold for determining sentence, obtains each in input text using the good term vector table of pre-training The vectorization of individual word is represented, forms continuous and dense real number vector matrix；
Step 2, the sentence term vector for input, pass through the information above of positive each word of LSTM e-learnings respectively With the context information of reverse each word of LSTM e-learnings, and the result of study is subjected to series connection merging, so that will bag Sentence term vector containing semantic information represents to be converted into while including the expression of semantic and contextual information；
Step 3, the term vector exported respectively to two-way LSTM networks using multiple different in width, the nuclear matrix comprising different weights Matrix carries out two-dimensional convolution computing, extracts local convolution feature, and generate the local convolution eigenmatrix of multilayer；
Step 4, using one-dimensional maximum pond algorithm down-sampling is carried out to the local convolution eigenmatrix of multilayer, obtain many of sentence Layer global characteristics matrix, and result is subjected to series connection merging；
Step 5, the long-term dependence between sentence local feature that learnt respectively using the LSTM networks of two opposite directions, and will Last learning outcome is exported；
Step 6, the output result of step 5 is first passed through to a full connection hidden layer, then again by one softmax layers to sentence Classification be predicted.
2. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：A kind of shot and long term memory network and the file classification method of convolutional neural networks of mixing is in a multilayer Completed in neutral net, the step 1 is completed in first layer input layer, step 2 is completed in the second layer is two-way LSTM layers, Step 3 is completed in CNN layers of third layer, and step 4 is completed in the 4th layer of pond layer, and step 5 is in layer 5 is two-way LSTM layers Complete, step 6 is completed in layer 6 output layer.
3. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 2, It is characterized in that：Two-way LSTM layers of the second layer is used for the contextual information for learning to be originally inputted the word of each in sentence, and Exported after the learning outcome of each word is connected, it is upper between sentence characteristics after the two-way LSTM layers of study convolution of layer 5 Context information, and only export the learning outcome of final step.
4. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：It is described pretreatment is carried out to sentence to include punctuation mark filtering, abbreviation polishing, deletion space, right in step 1 Sentence carries out participle and forbidden character filtering.
5. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：The step 3 is local feature learning process, passes through the two-dimensional convolution window and convolution of multiple different word step-lengths Term vector of the verification comprising contextual information is learnt, so as to obtain varigrained phrase information.
6. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：The step 4 is sampling and reduction process, by one-dimensional maximum pond algorithm to the local convolution feature of multilayer Matrix carries out down-sampling, obtains most important characteristic value in the window of the pond of each in sentence, and be used as the spy in local window Levy expression.
7. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：The step 5 learns for the context of local feature, above and below between two-way LSTM study local features Literary information, and the learning outcome of last term vector is exported, while forming the one-dimensional output of fixed dimension.
8. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：The step 6 exports for classification, by the multilayer perceptron connected entirely a progress classification judgement, and according to The probability distribution situation on taxonomic hierarchies is specified to obtain final output.
9. a kind of file classification method for mixing shot and long term memory network and convolutional neural networks according to claim 1, It is characterized in that：The step 6 is completed in the multilayer perceptron of two layers, including a full connection hidden layer and one Softmax layers, the output result of step 6 is the prediction classification of correspondence text.
CN201710257132.6A 2017-04-19 2017-04-19 A kind of file classification method mixing shot and long term memory network and convolutional neural networks Expired - Fee Related CN107169035B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710257132.6A CN107169035B (en) 2017-04-19 2017-04-19 A kind of file classification method mixing shot and long term memory network and convolutional neural networks

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710257132.6A CN107169035B (en) 2017-04-19 2017-04-19 A kind of file classification method mixing shot and long term memory network and convolutional neural networks

Publications (2)

Publication Number Publication Date
CN107169035A true CN107169035A (en) 2017-09-15
CN107169035B CN107169035B (en) 2019-10-18

Family

ID=59812256

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710257132.6A Expired - Fee Related CN107169035B (en) 2017-04-19 2017-04-19 A kind of file classification method mixing shot and long term memory network and convolutional neural networks

Country Status (1)

Country Link
CN (1) CN107169035B (en)

Cited By (69)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107679199A (en) * 2017-10-11 2018-02-09 北京邮电大学 A kind of external the Chinese text readability analysis method based on depth local feature
CN107832400A (en) * 2017-11-01 2018-03-23 山东大学 A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification
CN107908620A (en) * 2017-11-15 2018-04-13 珠海金山网络游戏科技有限公司 A kind of method and apparatus based on job documentation anticipation user's occupation
CN108334499A (en) * 2018-02-08 2018-07-27 海南云江科技有限公司 A kind of text label tagging equipment, method and computing device
CN108376558A (en) * 2018-01-24 2018-08-07 复旦大学 A kind of multi-modal nuclear magnetic resonance image Case report no automatic generation method
CN108415972A (en) * 2018-02-08 2018-08-17 合肥工业大学 text emotion processing method
CN108415923A (en) * 2017-10-18 2018-08-17 北京邮电大学 The intelligent interactive system of closed domain
CN108520320A (en) * 2018-03-30 2018-09-11 华中科技大学 A kind of equipment life prediction technique based on multiple shot and long term memory network and Empirical Bayes
CN108536825A (en) * 2018-04-10 2018-09-14 苏州市中地行信息技术有限公司 A method of whether identification source of houses data repeat
CN108595409A (en) * 2018-03-16 2018-09-28 上海大学 A kind of requirement documents based on neural network and service document matches method
CN108595429A (en) * 2018-04-25 2018-09-28 杭州闪捷信息科技股份有限公司 The method for carrying out Text character extraction based on depth convolutional neural networks
CN108595440A (en) * 2018-05-11 2018-09-28 厦门市美亚柏科信息股份有限公司 Short text content categorizing method and system
CN108614815A (en) * 2018-05-07 2018-10-02 华东师范大学 Sentence exchange method and device
CN108710651A (en) * 2018-05-08 2018-10-26 华南理工大学 A kind of large scale customer complaint data automatic classification method
CN108717439A (en) * 2018-05-16 2018-10-30 哈尔滨理工大学 A kind of Chinese Text Categorization merged based on attention mechanism and characteristic strengthening
CN108804591A (en) * 2018-05-28 2018-11-13 杭州依图医疗技术有限公司 A kind of file classification method and device of case history text
CN108829737A (en) * 2018-05-21 2018-11-16 浙江大学 Text combined crosswise classification method based on two-way shot and long term memory network
CN108874776A (en) * 2018-06-11 2018-11-23 北京奇艺世纪科技有限公司 A kind of recognition methods of rubbish text and device
CN108961816A (en) * 2018-07-19 2018-12-07 泰华智慧产业集团股份有限公司 Road parking berth prediction technique based on optimization LSTM model
CN108984745A (en) * 2018-07-16 2018-12-11 福州大学 A kind of neural network file classification method merging more knowledge mappings
CN109033413A (en) * 2018-03-12 2018-12-18 上海大学 A kind of requirement documents neural network based and service document matches method
CN109062996A (en) * 2018-07-05 2018-12-21 贵州威爱教育科技有限公司 A kind of management method and system of cloud file
CN109086892A (en) * 2018-06-15 2018-12-25 中山大学 It is a kind of based on the visual problem inference pattern and system that typically rely on tree
CN109101552A (en) * 2018-07-10 2018-12-28 东南大学 A kind of fishing website URL detection method based on deep learning
CN109213896A (en) * 2018-08-06 2019-01-15 杭州电子科技大学 Underwater video abstraction generating method based on shot and long term memory network intensified learning
CN109241284A (en) * 2018-08-27 2019-01-18 中国人民解放军国防科技大学 Document classification method and device
CN109271521A (en) * 2018-11-16 2019-01-25 北京九狐时代智能科技有限公司 A kind of file classification method and device
CN109271537A (en) * 2018-08-10 2019-01-25 北京大学 A kind of text based on distillation study is to image generating method and system
CN109308355A (en) * 2018-09-17 2019-02-05 清华大学 Legal decision prediction of result method and device
CN109359198A (en) * 2018-12-04 2019-02-19 北京容联易通信息技术有限公司 A kind of file classification method and device
CN109508377A (en) * 2018-11-26 2019-03-22 南京云思创智信息科技有限公司 Text feature, device, chat robots and storage medium based on Fusion Model
CN109508811A (en) * 2018-09-30 2019-03-22 中冶华天工程技术有限公司 Parameter prediction method is discharged based on principal component analysis and the sewage treatment of shot and long term memory network
CN109542585A (en) * 2018-11-14 2019-03-29 山东大学 A kind of Virtual Machine Worker load predicting method for supporting irregular time interval
CN109582794A (en) * 2018-11-29 2019-04-05 南京信息工程大学 Long article classification method based on deep learning
WO2019080864A1 (en) * 2017-10-27 2019-05-02 阿里巴巴集团控股有限公司 Semantic encoding method and device for text
CN109726268A (en) * 2018-08-29 2019-05-07 中国人民解放军国防科技大学 Text representation method and device based on hierarchical neural network
CN109743732A (en) * 2018-12-20 2019-05-10 重庆邮电大学 Refuse messages method of discrimination based on improved CNN-LSTM
CN109815456A (en) * 2019-02-13 2019-05-28 北京航空航天大学 A method of it is compressed based on term vector memory space of the character to coding
CN109840279A (en) * 2019-01-10 2019-06-04 山东亿云信息技术有限公司 File classification method based on convolution loop neural network
CN109902293A (en) * 2019-01-30 2019-06-18 华南理工大学 A kind of file classification method based on part with global mutually attention mechanism
CN109902301A (en) * 2019-02-26 2019-06-18 广东工业大学 Relation inference method, device and equipment based on deep neural network
CN109918503A (en) * 2019-01-29 2019-06-21 华南理工大学 The slot fill method of semantic feature is extracted from attention mechanism based on dynamic window
CN109947932A (en) * 2017-10-27 2019-06-28 中移(苏州)软件技术有限公司 A kind of pushed information classification method and system
CN109992781A (en) * 2019-04-02 2019-07-09 腾讯科技（深圳）有限公司 Processing, device, storage medium and the processor of text feature
CN110020431A (en) * 2019-03-06 2019-07-16 平安科技（深圳）有限公司 Feature extracting method, device, computer equipment and the storage medium of text information
CN110019784A (en) * 2017-09-29 2019-07-16 北京国双科技有限公司 A kind of file classification method and device
CN110046253A (en) * 2019-04-10 2019-07-23 广州大学 A kind of prediction technique of language conflict
CN110059192A (en) * 2019-05-15 2019-07-26 北京信息科技大学 Character level file classification method based on five codes
CN110083832A (en) * 2019-04-17 2019-08-02 北大方正集团有限公司 Recognition methods, device, equipment and the readable storage medium storing program for executing of article reprinting relationship
CN110196913A (en) * 2019-05-23 2019-09-03 北京邮电大学 Multiple entity relationship joint abstracting method and device based on text generation formula
CN110222953A (en) * 2018-12-29 2019-09-10 北京理工大学 A kind of power quality hybrid perturbation analysis method based on deep learning
CN110263152A (en) * 2019-05-07 2019-09-20 平安科技（深圳）有限公司 File classification method, system and computer equipment neural network based
CN110704890A (en) * 2019-08-12 2020-01-17 上海大学 Automatic text causal relationship extraction method fusing convolutional neural network and cyclic neural network
CN110781939A (en) * 2019-10-17 2020-02-11 中国铁塔股份有限公司 Method and device for detecting similar pictures and project management system
CN110837227A (en) * 2018-08-15 2020-02-25 格力电器（武汉）有限公司 Electric appliance control method and device
CN111126556A (en) * 2018-10-31 2020-05-08 百度在线网络技术（北京）有限公司 Training method and device of artificial neural network model
CN111371806A (en) * 2020-03-18 2020-07-03 北京邮电大学 Web attack detection method and device
CN111552808A (en) * 2020-04-20 2020-08-18 北京北大软件工程股份有限公司 Administrative illegal case law prediction method and tool based on convolutional neural network
CN111914085A (en) * 2020-06-18 2020-11-10 华南理工大学 Text fine-grained emotion classification method, system, device and storage medium
CN112052675A (en) * 2020-08-21 2020-12-08 北京邮电大学 Method and device for detecting sensitive information of unstructured text
WO2021004118A1 (en) * 2019-07-05 2021-01-14 深圳壹账通智能科技有限公司 Correlation value determination method and apparatus
CN112434156A (en) * 2020-11-02 2021-03-02 浙江大有实业有限公司杭州科技发展分公司 Power grid operation warning method and device based on mixed text classification model
US11010560B2 (en) 2018-11-08 2021-05-18 International Business Machines Corporation Multi-resolution convolutional neural networks for sequence modeling
CN112883708A (en) * 2021-02-25 2021-06-01 哈尔滨工业大学 Text inclusion recognition method based on 2D-LSTM
CN113780610A (en) * 2020-12-02 2021-12-10 北京沃东天骏信息技术有限公司 Customer service portrait construction method and device
WO2022227211A1 (en) * 2021-04-30 2022-11-03 平安科技（深圳）有限公司 Bert-based multi-intention recognition method for discourse, and device and readable storage medium
CN115563286A (en) * 2022-11-10 2023-01-03 东北农业大学 Knowledge-driven milk cow disease text classification method
CN116308464A (en) * 2023-05-11 2023-06-23 广州钛动科技股份有限公司 Target client acquisition system and method
CN116721361A (en) * 2023-06-09 2023-09-08 中国测绘科学研究院 Wetland remote sensing extraction method compatible with space-time discontinuous images

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104572892A (en) * 2014-12-24 2015-04-29 中国科学院自动化研究所 Text classification method based on cyclic convolution network
CN104834747A (en) * 2015-05-25 2015-08-12 中国科学院自动化研究所 Short text classification method based on convolution neutral network
US20170032221A1 (en) * 2015-07-29 2017-02-02 Htc Corporation Method, electronic apparatus, and computer readable medium of constructing classifier for disease detection
CN106547735A (en) * 2016-10-25 2017-03-29 复旦大学 The structure and using method of the dynamic word or word vector based on the context-aware of deep learning

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104572892A (en) * 2014-12-24 2015-04-29 中国科学院自动化研究所 Text classification method based on cyclic convolution network
CN104834747A (en) * 2015-05-25 2015-08-12 中国科学院自动化研究所 Short text classification method based on convolution neutral network
US20170032221A1 (en) * 2015-07-29 2017-02-02 Htc Corporation Method, electronic apparatus, and computer readable medium of constructing classifier for disease detection
CN106547735A (en) * 2016-10-25 2017-03-29 复旦大学 The structure and using method of the dynamic word or word vector based on the context-aware of deep learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
MINLIE HUANG: ""Modeling Rich Contexts for Sentiment Classification with LSTM"", 《ARXIV PREPRINT ARXIV:1605.01478》 *
黄磊等: ""基于递归神经网络的文本分类研究"", 《北京化工大学学报》 *

Cited By (98)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110019784A (en) * 2017-09-29 2019-07-16 北京国双科技有限公司 A kind of file classification method and device
CN107679199A (en) * 2017-10-11 2018-02-09 北京邮电大学 A kind of external the Chinese text readability analysis method based on depth local feature
CN108415923A (en) * 2017-10-18 2018-08-17 北京邮电大学 The intelligent interactive system of closed domain
CN108415923B (en) * 2017-10-18 2020-12-11 北京邮电大学 Intelligent man-machine conversation system of closed domain
CN109947932A (en) * 2017-10-27 2019-06-28 中移(苏州)软件技术有限公司 A kind of pushed information classification method and system
CN110019793A (en) * 2017-10-27 2019-07-16 阿里巴巴集团控股有限公司 A kind of text semantic coding method and device
WO2019080864A1 (en) * 2017-10-27 2019-05-02 阿里巴巴集团控股有限公司 Semantic encoding method and device for text
JP2021501390A (en) * 2017-10-27 2021-01-14 アリババ・グループ・ホールディング・リミテッドＡｌｉｂａｂａ Ｇｒｏｕｐ Ｈｏｌｄｉｎｇ Ｌｉｍｉｔｅｄ Text Semantic Coding Methods and Devices
CN107832400B (en) * 2017-11-01 2019-04-16 山东大学 A kind of method that location-based LSTM and CNN conjunctive model carries out relationship classification
CN107832400A (en) * 2017-11-01 2018-03-23 山东大学 A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification
CN107908620A (en) * 2017-11-15 2018-04-13 珠海金山网络游戏科技有限公司 A kind of method and apparatus based on job documentation anticipation user's occupation
CN108376558A (en) * 2018-01-24 2018-08-07 复旦大学 A kind of multi-modal nuclear magnetic resonance image Case report no automatic generation method
CN108376558B (en) * 2018-01-24 2021-08-20 复旦大学 Automatic generation method for multi-modal nuclear magnetic resonance image medical record report
CN108334499A (en) * 2018-02-08 2018-07-27 海南云江科技有限公司 A kind of text label tagging equipment, method and computing device
CN108415972A (en) * 2018-02-08 2018-08-17 合肥工业大学 text emotion processing method
CN108334499B (en) * 2018-02-08 2022-03-18 海南云江科技有限公司 Text label labeling device and method and computing device
CN109033413B (en) * 2018-03-12 2022-12-23 上海大学 Neural network-based demand document and service document matching method
CN109033413A (en) * 2018-03-12 2018-12-18 上海大学 A kind of requirement documents neural network based and service document matches method
CN108595409A (en) * 2018-03-16 2018-09-28 上海大学 A kind of requirement documents based on neural network and service document matches method
CN108520320A (en) * 2018-03-30 2018-09-11 华中科技大学 A kind of equipment life prediction technique based on multiple shot and long term memory network and Empirical Bayes
CN108536825A (en) * 2018-04-10 2018-09-14 苏州市中地行信息技术有限公司 A method of whether identification source of houses data repeat
CN108595429A (en) * 2018-04-25 2018-09-28 杭州闪捷信息科技股份有限公司 The method for carrying out Text character extraction based on depth convolutional neural networks
CN108614815A (en) * 2018-05-07 2018-10-02 华东师范大学 Sentence exchange method and device
CN108710651A (en) * 2018-05-08 2018-10-26 华南理工大学 A kind of large scale customer complaint data automatic classification method
CN108710651B (en) * 2018-05-08 2022-03-25 华南理工大学 Automatic classification method for large-scale customer complaint data
CN108595440A (en) * 2018-05-11 2018-09-28 厦门市美亚柏科信息股份有限公司 Short text content categorizing method and system
CN108595440B (en) * 2018-05-11 2022-03-18 厦门市美亚柏科信息股份有限公司 Short text content classification method and system
CN108717439A (en) * 2018-05-16 2018-10-30 哈尔滨理工大学 A kind of Chinese Text Categorization merged based on attention mechanism and characteristic strengthening
CN108829737B (en) * 2018-05-21 2021-11-05 浙江大学 Text cross combination classification method based on bidirectional long-short term memory network
CN108829737A (en) * 2018-05-21 2018-11-16 浙江大学 Text combined crosswise classification method based on two-way shot and long term memory network
CN108804591A (en) * 2018-05-28 2018-11-13 杭州依图医疗技术有限公司 A kind of file classification method and device of case history text
CN108874776A (en) * 2018-06-11 2018-11-23 北京奇艺世纪科技有限公司 A kind of recognition methods of rubbish text and device
CN108874776B (en) * 2018-06-11 2022-06-03 北京奇艺世纪科技有限公司 Junk text recognition method and device
CN109086892A (en) * 2018-06-15 2018-12-25 中山大学 It is a kind of based on the visual problem inference pattern and system that typically rely on tree
CN109086892B (en) * 2018-06-15 2022-02-18 中山大学 General dependency tree-based visual problem reasoning model and system
CN109062996A (en) * 2018-07-05 2018-12-21 贵州威爱教育科技有限公司 A kind of management method and system of cloud file
CN109101552B (en) * 2018-07-10 2022-01-28 东南大学 Phishing website URL detection method based on deep learning
CN109101552A (en) * 2018-07-10 2018-12-28 东南大学 A kind of fishing website URL detection method based on deep learning
CN108984745B (en) * 2018-07-16 2021-11-02 福州大学 Neural network text classification method fusing multiple knowledge maps
CN108984745A (en) * 2018-07-16 2018-12-11 福州大学 A kind of neural network file classification method merging more knowledge mappings
CN108961816A (en) * 2018-07-19 2018-12-07 泰华智慧产业集团股份有限公司 Road parking berth prediction technique based on optimization LSTM model
CN109213896B (en) * 2018-08-06 2021-06-01 杭州电子科技大学 Underwater video abstract generation method based on long-short term memory network reinforcement learning
CN109213896A (en) * 2018-08-06 2019-01-15 杭州电子科技大学 Underwater video abstraction generating method based on shot and long term memory network intensified learning
CN109271537A (en) * 2018-08-10 2019-01-25 北京大学 A kind of text based on distillation study is to image generating method and system
CN109271537B (en) * 2018-08-10 2021-11-23 北京大学 Text-to-image generation method and system based on distillation learning
CN110837227A (en) * 2018-08-15 2020-02-25 格力电器（武汉）有限公司 Electric appliance control method and device
CN109241284A (en) * 2018-08-27 2019-01-18 中国人民解放军国防科技大学 Document classification method and device
CN109726268A (en) * 2018-08-29 2019-05-07 中国人民解放军国防科技大学 Text representation method and device based on hierarchical neural network
CN109308355A (en) * 2018-09-17 2019-02-05 清华大学 Legal decision prediction of result method and device
CN109308355B (en) * 2018-09-17 2020-03-13 清华大学 Legal judgment result prediction method and device
CN109508811A (en) * 2018-09-30 2019-03-22 中冶华天工程技术有限公司 Parameter prediction method is discharged based on principal component analysis and the sewage treatment of shot and long term memory network
CN111126556B (en) * 2018-10-31 2023-07-25 百度在线网络技术（北京）有限公司 Training method and device for artificial neural network model
CN111126556A (en) * 2018-10-31 2020-05-08 百度在线网络技术（北京）有限公司 Training method and device of artificial neural network model
US11010560B2 (en) 2018-11-08 2021-05-18 International Business Machines Corporation Multi-resolution convolutional neural networks for sequence modeling
CN109542585A (en) * 2018-11-14 2019-03-29 山东大学 A kind of Virtual Machine Worker load predicting method for supporting irregular time interval
CN109542585B (en) * 2018-11-14 2020-06-16 山东大学 Virtual machine workload prediction method supporting irregular time intervals
CN109271521A (en) * 2018-11-16 2019-01-25 北京九狐时代智能科技有限公司 A kind of file classification method and device
CN109508377A (en) * 2018-11-26 2019-03-22 南京云思创智信息科技有限公司 Text feature, device, chat robots and storage medium based on Fusion Model
CN109582794A (en) * 2018-11-29 2019-04-05 南京信息工程大学 Long article classification method based on deep learning
CN109359198A (en) * 2018-12-04 2019-02-19 北京容联易通信息技术有限公司 A kind of file classification method and device
CN109743732B (en) * 2018-12-20 2022-05-10 重庆邮电大学 Junk short message distinguishing method based on improved CNN-LSTM
CN109743732A (en) * 2018-12-20 2019-05-10 重庆邮电大学 Refuse messages method of discrimination based on improved CNN-LSTM
CN110222953A (en) * 2018-12-29 2019-09-10 北京理工大学 A kind of power quality hybrid perturbation analysis method based on deep learning
CN109840279A (en) * 2019-01-10 2019-06-04 山东亿云信息技术有限公司 File classification method based on convolution loop neural network
CN109918503A (en) * 2019-01-29 2019-06-21 华南理工大学 The slot fill method of semantic feature is extracted from attention mechanism based on dynamic window
CN109918503B (en) * 2019-01-29 2020-12-22 华南理工大学 Groove filling method for extracting semantic features based on dynamic window self-attention mechanism
CN109902293A (en) * 2019-01-30 2019-06-18 华南理工大学 A kind of file classification method based on part with global mutually attention mechanism
CN109815456A (en) * 2019-02-13 2019-05-28 北京航空航天大学 A method of it is compressed based on term vector memory space of the character to coding
CN109902301B (en) * 2019-02-26 2023-02-10 广东工业大学 Deep neural network-based relationship reasoning method, device and equipment
CN109902301A (en) * 2019-02-26 2019-06-18 广东工业大学 Relation inference method, device and equipment based on deep neural network
CN110020431A (en) * 2019-03-06 2019-07-16 平安科技（深圳）有限公司 Feature extracting method, device, computer equipment and the storage medium of text information
CN109992781A (en) * 2019-04-02 2019-07-09 腾讯科技（深圳）有限公司 Processing, device, storage medium and the processor of text feature
CN110046253A (en) * 2019-04-10 2019-07-23 广州大学 A kind of prediction technique of language conflict
CN110046253B (en) * 2019-04-10 2022-01-04 广州大学 Language conflict prediction method
CN110083832A (en) * 2019-04-17 2019-08-02 北大方正集团有限公司 Recognition methods, device, equipment and the readable storage medium storing program for executing of article reprinting relationship
CN110083832B (en) * 2019-04-17 2020-12-29 北大方正集团有限公司 Article reprint relation identification method, device, equipment and readable storage medium
CN110263152B (en) * 2019-05-07 2024-04-09 平安科技（深圳）有限公司 Text classification method, system and computer equipment based on neural network
CN110263152A (en) * 2019-05-07 2019-09-20 平安科技（深圳）有限公司 File classification method, system and computer equipment neural network based
CN110059192A (en) * 2019-05-15 2019-07-26 北京信息科技大学 Character level file classification method based on five codes
CN110196913A (en) * 2019-05-23 2019-09-03 北京邮电大学 Multiple entity relationship joint abstracting method and device based on text generation formula
WO2021004118A1 (en) * 2019-07-05 2021-01-14 深圳壹账通智能科技有限公司 Correlation value determination method and apparatus
CN110704890A (en) * 2019-08-12 2020-01-17 上海大学 Automatic text causal relationship extraction method fusing convolutional neural network and cyclic neural network
CN110781939A (en) * 2019-10-17 2020-02-11 中国铁塔股份有限公司 Method and device for detecting similar pictures and project management system
CN111371806A (en) * 2020-03-18 2020-07-03 北京邮电大学 Web attack detection method and device
CN111552808A (en) * 2020-04-20 2020-08-18 北京北大软件工程股份有限公司 Administrative illegal case law prediction method and tool based on convolutional neural network
CN111914085A (en) * 2020-06-18 2020-11-10 华南理工大学 Text fine-grained emotion classification method, system, device and storage medium
CN111914085B (en) * 2020-06-18 2024-04-23 华南理工大学 Text fine granularity emotion classification method, system, device and storage medium
CN112052675A (en) * 2020-08-21 2020-12-08 北京邮电大学 Method and device for detecting sensitive information of unstructured text
CN112434156A (en) * 2020-11-02 2021-03-02 浙江大有实业有限公司杭州科技发展分公司 Power grid operation warning method and device based on mixed text classification model
CN113780610A (en) * 2020-12-02 2021-12-10 北京沃东天骏信息技术有限公司 Customer service portrait construction method and device
CN112883708A (en) * 2021-02-25 2021-06-01 哈尔滨工业大学 Text inclusion recognition method based on 2D-LSTM
WO2022227211A1 (en) * 2021-04-30 2022-11-03 平安科技（深圳）有限公司 Bert-based multi-intention recognition method for discourse, and device and readable storage medium
CN115563286A (en) * 2022-11-10 2023-01-03 东北农业大学 Knowledge-driven milk cow disease text classification method
CN115563286B (en) * 2022-11-10 2023-12-01 东北农业大学 Knowledge-driven dairy cow disease text classification method
CN116308464B (en) * 2023-05-11 2023-09-08 广州市沃钛移动科技有限公司 Target client acquisition system and method
CN116308464A (en) * 2023-05-11 2023-06-23 广州钛动科技股份有限公司 Target client acquisition system and method
CN116721361A (en) * 2023-06-09 2023-09-08 中国测绘科学研究院 Wetland remote sensing extraction method compatible with space-time discontinuous images
CN116721361B (en) * 2023-06-09 2024-01-02 中国测绘科学研究院 Wetland remote sensing extraction method compatible with space-time discontinuous images

Also Published As

Publication number Publication date
CN107169035B (en) 2019-10-18

Similar Documents

Publication Publication Date Title
CN107169035B (en) A kind of file classification method mixing shot and long term memory network and convolutional neural networks
CN110866117B (en) Short text classification method based on semantic enhancement and multi-level label embedding
CN107133213B (en) Method and system for automatically extracting text abstract based on algorithm
CN107066553B (en) Short text classification method based on convolutional neural network and random forest
Wang et al. Research on Web text classification algorithm based on improved CNN and SVM
CN107832400A (en) A kind of method that location-based LSTM and CNN conjunctive models carry out relation classification
CN109189925A (en) Term vector model based on mutual information and based on the file classification method of CNN
CN107818164A (en) A kind of intelligent answer method and its system
CN107918782A (en) A kind of method and system for the natural language for generating description picture material
CN110321563B (en) Text emotion analysis method based on hybrid supervision model
CN106599933A (en) Text emotion classification method based on the joint deep learning model
CN107193801A (en) A kind of short text characteristic optimization and sentiment analysis method based on depth belief network
CN110765260A (en) Information recommendation method based on convolutional neural network and joint attention mechanism
CN107451278A (en) Chinese Text Categorization based on more hidden layer extreme learning machines
CN111143563A (en) Text classification method based on integration of BERT, LSTM and CNN
CN110415071B (en) Automobile competitive product comparison method based on viewpoint mining analysis
CN111858878B (en) Method, system and storage medium for automatically extracting answer from natural language text
Zhang Research on text classification method based on LSTM neural network model
CN112287106A (en) Online comment emotion classification method based on dual-channel hybrid neural network
CN106570170A (en) Text classification and naming entity recognition integrated method and system based on depth cyclic neural network
CN116484262B (en) Textile equipment fault auxiliary processing method based on text classification
CN113220890A (en) Deep learning method combining news headlines and news long text contents based on pre-training
CN111651602A (en) Text classification method and system
CN114239585A (en) Biomedical nested named entity recognition method
CN114462420A (en) False news detection method based on feature fusion model

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
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20191018

CF01 Termination of patent right due to non-payment of annual fee