CN109165294A - Short text classification method based on Bayesian classification - Google Patents
Short text classification method based on Bayesian classification Download PDFInfo
- Publication number
- CN109165294A CN109165294A CN201810951636.2A CN201810951636A CN109165294A CN 109165294 A CN109165294 A CN 109165294A CN 201810951636 A CN201810951636 A CN 201810951636A CN 109165294 A CN109165294 A CN 109165294A
- Authority
- CN
- China
- Prior art keywords
- classification
- data
- short text
- text
- probability
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a short text classification method based on Bayesian classification, which relates to the field of smart cities and electronic government affairs, and comprises the following steps: (1) preprocessing data and labeling categories; (2) completing word segmentation and incremental feature vector extraction of short text data, and mainly comprising the following two core steps; (3) establishing a short text classification model based on Bayes; (4) dividing the processed data set into a training set and a testing set, carrying out classification model training, and carrying out model optimization according to the result of the training set; (5) according to the trained model, short text data of unknown classes are input, the probability that the current input text belongs to each class is output, the class with the highest probability is selected as the result of the final classification class, and the short text classification method based on the Bayesian classification can effectively, intelligently and automatically classify the short text content.
Description
Technical field
The present invention relates to smart cities and E-Government field, and in particular to a kind of short essay one's duty based on Bayes's classification
Class method.
Background technique:
With the development of mobile Internet and social networks, the rise of the social softwares such as microblogging, wechat, company and portion of government
Door is also gradually established connection using social software, is linked up.Issue frequency height, short and small content of text is mobile social media
Feature, the scale of short text content are also being skyrocketed through.In search engine, intelligent customer service and public sentiment monitoring field, short text
It is studied emphasis.In face of so huge and constantly incremental netizen's quantity, from various phenomenon description, personal letter,
In the imperfect text informations such as comment, useful information is extracted, seems particularly important to policymaker such as media, governments.Manually
Therefore, how efficiently, intelligence processing huge large-scale short text classification such as extracts at the inefficiency, can not usually efficiently accomplish task,
Can, important in inhibiting of the effective classification to the construction for promoting E-Government automatically be carried out to short text content.
For the technology of existing text classification mainly from the representative degree of keyword, i.e. popularity proposes the similar methods such as weight
To carry out the design of core classification algorithm;For example, in existing literature " a kind of file classification method based on cluster word insertion ", it is main
If k- mean algorithm is applied on the word vector of document, the cluster set of fixed size, the mass center of each cluster are obtained
It is interpreted a super word insertion, each insertion word in text collection is assigned to nearest cluster centers.Each cluster
Mass center be interpreted a super word insertion, each insertion word in text collection is assigned to nearest cluster centers.Often
A text is represented as a super word insertion packet, calculates each super word and is embedded in the frequency in respective text, that is, obtains
Obtain the type of text.
Above-mentioned short text classification method is analyzed it is found that the selection of keyword affects classifying quality, needs to consider keyword
Quantity and feature popularity, and short text classification in, short text characteristic key words are few, during actual classification, close
Keyword is difficult to the inherent meaning of effective expression short text, and being easy to produce a text, there are the results of multiple class categories;In addition,
Semantic information in short text also affect classification as a result, and in the prior art extract characteristic key words method to long text
Classification there is preferable effect, and short text is difficult to effectively classify
Such as application No. is CN201710216502.1 to disclose a kind of text classifier obtained for automatic marking corpus
Method and text classifier, this method includes determining concept set, with general in the corresponding concept keyword set of each concept
It reads keyword and match simultaneously automatic marking processing to un-annotated data text;For each concept, when the corresponding mark of the concept
When amount of text meets threshold condition in note corpus text collection, then corresponding text classification mould is trained to the concept
Type, obtains corresponding text classifier, finally obtains the text corresponding with the concept that all amount of text meet threshold condition
Classifier set.This kind of algorithm structure has universality, can neatly change classification system, save calculating time and resource,
And the present invention provides a small amount of initial corpus text, and automatic marking, without artificial mark, further save the time and
Cost, but this kind of classification method is it is not disclosed how make the higher technical solution of its accuracy by independently training.
Such as application No. is CN201710882685.0 disclose a kind of method for establishing textual classification model and text classification,
Device, method for building up include: acquisition training sample;Corresponding moment of a vector is obtained after carrying out word cutting to text based on entity dictionary
Battle array;Using the vector matrix of text and the classification of text, the first disaggregated model of training and the second disaggregated model;In training process
In, the loss function of textual classification model is obtained using the loss function of the first disaggregated model and the second disaggregated model, and utilize
The loss function of textual classification model obtains by the first and second disaggregated model structures the first and second disaggregated model adjusting parameters
At textual classification model.The method of text classification includes: to obtain text to be sorted;Text is cut based on entity dictionary
The corresponding vector matrix of text is obtained after word;Vector matrix is inputted into textual classification model, according to the output of textual classification model,
The classification results of the text are obtained, but this kind of classification method by independently training it is not disclosed how keep its accuracy higher
Technical solution.
Summary of the invention
The purpose of the present invention is to provide a kind of short text classification methods based on Bayes's classification, to solve the prior art
In caused above-mentioned defects.
A kind of short text classification method based on Bayes's classification, which is characterized in that this method includes following steps:
(1) data prediction and classification mark:
Step 1: the history short text data reported is extracted, and routine data cleaning, data integration etc. are carried out to data
Reason improves the quality of data;
Step 2: to the data after preliminary cleaning are completed, the processed short text of history has been accomplished manually classification mark, to working as
Preceding untreated partial data carries out artificial classification mark, completes process of data preprocessing;
(2) participle and increment feature vector for completing short text data extract, and are broadly divided into following two core procedure:
Step 1: the three-party library Jieba participle based on Python segments the short text content after cleaning;
Step 2: proposing increment feature vector, and TF-IDF combined to carry out keyword extraction, if keyword is very few,
It directly uses and all segments phrase as final sorting parameter input;
(3) the short text disaggregated model based on Bayes is established;
(4) training set and test set are divided into processed data acquisition system, carry out disaggregated model training, and according to training
The result of collection carries out the optimization of model;
(5) it according to trained model, the short text data of unknown input classification, exports currently to input text and belonging to
The probability of each classification chooses result of the classification of maximum probability as final classification classification.
Preferably, the data prediction includes following four step:
Step 1: carrying out cleaning classification for initial data, and text is divided into three classifications using Kettle, is major class respectively
Serial number, group serial number and text;
Step 2: the data handled well are stored in database;
Step 3: the content i.e. plain text of third field are segmented using Jiaba participle;
Step 4: the every row of the word divided is left by three word deposit databases according to part of speech.
Preferably, the increment feature vector sum TF-IDF Feature Words extraction method carry out the extraction of characteristic key words include with
Lower two steps:
Step 1: note B=(B1,B2,...,Bu) be the Feature Words composition extracted from text feature vector, u herein
Value be it is smaller, such as 3 or 4, such as rubbish reports information, there is a floating on water surface in the position of rubbish, and in greenbelt, road surface will
The word for describing the distributing position of rubbish is summarised as a new Feature Words Bu+1, it gives and names, and so on, work as u=5,
6 ..., m just obtains increment feature vector B=(B1,B2,...,Bm);
Step 2: if the TF high of some word or phrase frequency of occurrences in an article, and in other articles very
It is few to occur, then it is assumed that this word or phrase have good class discrimination ability, are adapted to classify, the feature extraction of TF-IDF
Function are as follows: f (w)=TF (w) × IDF (w) × log [N*n (w)+1] completes feature to short text content according to above-mentioned formula and closes
Keyword extracts, firstly, the TF value of Feature Words w is denoted asTF(w), often characteristic item frequency TF is risen in conjunction with anti-document frequency IDF
To use;Then IDF (w)=log [N*n (w)+1] is calculated, N is text sum, and n (w) is the textual data comprising w.
Preferably, to the short text sample record of input, B=(B1, B2 ..., Bm) be extract feature vector, C1,
C2 ..., Cn is n classification results;P (Ci | B), i=1,2 ..., n indicates that text to be sorted belongs to i-th of classification results
Probability P (Bj| Ci), j=1,2 ..., m, i=1,2 ..., n indicate that j-th of Feature Words belongs to the probability of the i-th class, specifically counting
It is lower shown based on Bayesian formula in calculation:
When classify new text when, it is only necessary to new sample is sentenced probability by the value for calculating P in n classification (Ci | B)
It is worth in maximum class, wherein probability P (B) is the constant unrelated with classification, further according to feature vector B=(B1,B2,...,Bm) each
Independence between a Feature Words, above-mentioned calculation formula can simplify are as follows:
Preferably, according to the model of foundation, the classification ownership of unknown short text information is calculated, if N is the sample of prediction
Sum, Cou (Ci) indicate i-th of classification counting in the sample, then P (Ci)=Cou (Ci)/N, Cou (Bij) indicate i-th point
In class, the number of j-th of Feature Words, then P (Bj|Ci)=Cou (Bij)/Cou(Ci), belong to finally, calculating sample to be sorted
The probability of each classification obtains maximum probability
The present invention has the advantages that the short text classification method based on Bayes's classification is somebody's turn to do, according to the short essay of reporting of user
Classified after this content analysis and be distributed to service unit, for the short text assorting process of core, first to source data into
Row data cleansing, the regular processing such as integrated, and extraction section short essay data are as training data, according to the demand of classification to extraction
Data carry out classification annotation;Then, the three-party library Jieba participle based on Python divides the short text content after cleaning
Word, and keyword is extracted based on TF-IDF, it is contemplated that short text content is few, and therefore, the keyword that TF-IDF is extracted is as pattra leaves
Reference before this classification model construction, if the keyword extracted is very few, the phrase after the short text that then be used directly participle carries out classification and builds
Mould establishes disaggregated model based on Bayesian formula, and adjust correlation model according to above-mentioned steps, until the precision of class test
Until tending towards stability.
Detailed description of the invention
Fig. 1 is flow chart of the method for the present invention.
Fig. 2 is the flow chart of data processing in the present invention.
Specific embodiment
To be easy to understand the technical means, the creative features, the aims and the efficiencies achieved by the present invention, below with reference to
Specific embodiment, the present invention is further explained.
As depicted in figs. 1 and 2, a kind of short text classification method based on Bayes's classification, which is characterized in that this method packet
Include following steps:
(1) data prediction and classification mark:
Step 1: the history short text data reported is extracted, and routine data cleaning, data integration etc. are carried out to data
Reason improves the quality of data;
Step 2: to the data after preliminary cleaning are completed, the processed short text of history has been accomplished manually classification mark, to working as
Preceding untreated partial data carries out artificial classification mark, completes process of data preprocessing;
(2) participle and increment feature vector for completing short text data extract, and are broadly divided into following two core procedure:
Step 1: the three-party library Jieba participle based on Python segments the short text content after cleaning;
Step 2: proposing increment feature vector, and TF-IDF combined to carry out keyword extraction, if keyword is very few,
It directly uses and all segments phrase as final sorting parameter input;
(3) the short text disaggregated model based on Bayes is established;
(4) training set and test set are divided into processed data acquisition system, carry out disaggregated model training, and according to training
The result of collection carries out the optimization of model;
(5) it according to trained model, the short text data of unknown input classification, exports currently to input text and belonging to
The probability of each classification chooses result of the classification of maximum probability as final classification classification.
It is worth noting that, the data prediction includes following four step:
Step 1: carrying out cleaning classification for initial data, and text is divided into three classifications using Kettle, is major class respectively
Serial number, group serial number and text;
Step 2: the data handled well are stored in database;
Step 3: the content i.e. plain text of third field are segmented using Jiaba participle;
Step 4: the every row of the word divided is left by three word deposit databases according to part of speech.
In the present embodiment, the increment feature vector sum TF-IDF Feature Words extraction method carries out the extraction of characteristic key words
Including following two step:
Step 1: note B=(B1,B2,...,Bu) be the Feature Words composition extracted from text feature vector, u herein
Value be it is smaller, such as 3 or 4, such as rubbish reports information, there is a floating on water surface in the position of rubbish, and in greenbelt, road surface will
The word for describing the distributing position of rubbish is summarised as a new Feature Words Bu+1, it gives and names, and so on, work as u=5,
6 ..., m just obtains increment feature vector B=(B1,B2,...,Bm);
Step 2: if the TF high of some word or phrase frequency of occurrences in an article, and in other articles very
It is few to occur, then it is assumed that this word or phrase have good class discrimination ability, are adapted to classify, the feature extraction of TF-IDF
Function are as follows: f (w)=TF (w) × IDF (w) × log [N*n (w)+1] completes feature to short text content according to above-mentioned formula and closes
Keyword extracts, firstly, the TF value of Feature Words w is denoted as TF (w), often by characteristic item frequency TF in conjunction with anti-document frequency IDF
Get up to use;Then IDF (w)=log [N*n (w)+1] is calculated, N is text sum, and n (w) is the textual data comprising w.
In the present embodiment, to the short text sample record of input, B=(B1, B2,...,Bm) it is the feature vector extracted,
C1,C2,...,CnFor n classification results;P(Ci| B), i=1,2 ..., n indicates that text to be sorted belongs to i-th of classification results
Probability P (Bj|Ci), j=1,2 ..., m, i=1,2 ..., n indicate that j-th of Feature Words belongs to the probability of the i-th class, specific
It is lower shown based on Bayesian formula in calculating:
When the new text of classification, it is only necessary to calculate P (C in n classificationi| B) value, new sample is sentenced into probability
It is worth in maximum class, wherein probability P (B) is the constant unrelated with classification, further according to feature vector B=(B1,B2,...,Bm) each
Independence between a Feature Words, above-mentioned calculation formula can simplify are as follows:
In addition, according to the model of foundation, the classification ownership of unknown short text information is calculated, if N be that the sample predicted is total
Number, Cou (Ci) indicate i-th of classification counting in the sample, then P (Ci)=Cou (Ci)/N, Cou (Bij) indicate i-th of classification
In, the number of j-th of Feature Words, then P (Bj|Ci)=Cou (Bij)/Cou(Ci), belong to often finally, calculating sample to be sorted
The probability of a classification obtains maximum probability
Based on above-mentioned, should short text classification method based on Bayes's classification, this method includes following steps: (1) number
Data preprocess and classification mark;(2) participle and increment feature vector for completing short text data extract, and are broadly divided into following two
Core procedure;(3) the short text disaggregated model based on Bayes is established;(4) training set is divided into processed data acquisition system
And test set, disaggregated model training is carried out, and the optimization of model is carried out according to the result of training set;(5) according to trained
Model, the short text data of unknown input classification export currently to input the probability that text belongs to each classification, choose probability most
Big classification is as final classification classification as a result, being classified and being distributed to according to after the short text content analysis of reporting of user
Service unit carries out data cleansing, the regular processing such as integrated to source data first for the short text assorting process of core, and
Extraction section short essay data carry out classification annotation as training data, according to data of the demand of classification to extraction;Then, it is based on
The three-party library Jieba participle of Python segments the short text content after cleaning, and extracts keyword based on TF-IDF, examines
It is few to consider short text content, therefore, the keyword that TF-IDF is extracted is as the reference before Bayes's classification modeling, if the pass extracted
Keyword is very few, and the phrase after the short text that then be used directly participle carries out classification model construction, according to above-mentioned steps, is based on Bayesian formula
Disaggregated model is established, and adjusts correlation model, until the precision of class test tends towards stability.
As known by the technical knowledge, the present invention can pass through the embodiment party of other essence without departing from its spirit or essential feature
Case is realized.Therefore, embodiment disclosed above, in all respects are merely illustrative, not the only.Institute
Have within the scope of the present invention or is included in the invention in the change being equal in the scope of the present invention.
Claims (5)
1. a kind of short text classification method based on Bayes's classification, which is characterized in that this method includes following steps:
(1) data prediction and classification mark:
Step 1: extracting the history short text data reported, and carry out routine data cleaning to data, and data integration etc. is handled,
Improve the quality of data;
Step 2: to the data after preliminary cleaning are completed, the processed short text of history has been accomplished manually classification mark, to currently not
The partial data of processing carries out artificial classification mark, completes process of data preprocessing;
(2) participle and increment feature vector for completing short text data extract, and are broadly divided into following two core procedure:
Step 1: the three-party library Jieba participle based on Python segments the short text content after cleaning;
Step 2: increment feature vector is proposed, and TF-IDF is combined to carry out keyword extraction, if keyword is very few, directly
Use and all segments phrase as final sorting parameter input;
(3) the short text disaggregated model based on Bayes is established;
(4) training set and test set are divided into processed data acquisition system, carry out disaggregated model training, and according to training set
As a result the optimization of model is carried out;
(5) according to trained model, the short text data of unknown input classification, export for currently input text belong to it is each
The probability of classification chooses result of the classification of maximum probability as final classification classification.
2. a kind of short text classification method based on Bayes's classification according to claim 1, it is characterised in that: the number
Data preprocess includes following four step:
Step 1: carrying out cleaning classification for initial data, and text is divided into three classifications using Kettle, is major class sequence respectively
Number, group serial number and text;
Step 2: the data handled well are stored in database;
Step 3: the content i.e. plain text of third field are segmented using Jiaba participle;
Step 4: the every row of the word divided is left by three word deposit databases according to part of speech.
3. a kind of short text classification method based on Bayes's classification according to claim 1, it is characterised in that: the increasing
The extraction that measure feature vector sum TF-IDF Feature Words extraction method carries out characteristic key words includes following two step:
Step 1: note B=(B1,B2,...,Bu) be the Feature Words composition extracted from text feature vector, the value of u herein
Be it is smaller, such as 3 or 4, such as rubbish reports information, there is a floating on water surface in the position of rubbish, and in greenbelt, road surface will be described
The word of the distributing position of rubbish is summarised as a new Feature Words Bu+1, it gives and names, and so on, work as u=5,6 ..., m
Just increment feature vector B=(B is obtained1,B2,...,Bm);
Step 2: if the TF high of some word or phrase frequency of occurrences in an article, and seldom go out in other articles
It is existing, then it is assumed that this word or phrase have good class discrimination ability, are adapted to classify, the feature extraction function of TF-IDF
Are as follows: f (w)=TF (w) × IDF (w) × log [N*n (w)+1] completes characteristic key words to short text content according to above-mentioned formula
It extracts, firstly, the TF value of Feature Words w is denoted as TF (w), often combines characteristic item frequency TF with anti-document frequency IDF
It uses;Then IDF (w)=log [N*n (w)+1] is calculated, N is text sum, and n (w) is the textual data comprising w.
4. a kind of short text classification method based on Bayes's classification according to claim 3, it is characterised in that: to input
Short text sample record, B=(B1,B2,...,Bm) it is the feature vector extracted, C1,C2,...,CnFor n classification results;P
(Ci| B), i=1,2 ..., n indicates that text to be sorted belongs to the probability P (B of i-th of classification resultsj|Ci), j=1,2 ..., m,
I=1,2 ..., n indicate that j-th of Feature Words belongs to the probability of the i-th class, are lower institute based on Bayesian formula in specific calculate
Show:
When the new text of classification, it is only necessary to calculate P (C in n classificationi| B) value, by new sample sentence probability value maximum
Class in, wherein probability P (B) is the constant unrelated with classification, further according to feature vector B=(B1,B2,...,Bm) each feature
Independence between word, above-mentioned calculation formula can simplify are as follows:
5. a kind of short text classification method based on Bayes's classification according to claim 1, it is characterised in that: according to building
Vertical model calculates the classification ownership of unknown short text information, if N is the total sample number of prediction, Cou (Ci) indicate i-th
Classification counting in the sample, then P (Ci)=Cou (Ci)/N, Cou (Bij) indicate in i-th of classification, of j-th of Feature Words
It counts, then P (Bj|Ci)=Cou (Bij)/Cou(Ci), finally, calculating the probability that sample to be sorted belongs to each classification, obtain most
Big probability
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810951636.2A CN109165294B (en) | 2018-08-21 | 2018-08-21 | Short text classification method based on Bayesian classification |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810951636.2A CN109165294B (en) | 2018-08-21 | 2018-08-21 | Short text classification method based on Bayesian classification |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109165294A true CN109165294A (en) | 2019-01-08 |
CN109165294B CN109165294B (en) | 2021-09-24 |
Family
ID=64896189
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810951636.2A Active CN109165294B (en) | 2018-08-21 | 2018-08-21 | Short text classification method based on Bayesian classification |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109165294B (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110619363A (en) * | 2019-09-17 | 2019-12-27 | 陕西优百信息技术有限公司 | Classification method for subclass names corresponding to long description of material data |
CN111159414A (en) * | 2020-04-02 | 2020-05-15 | 成都数联铭品科技有限公司 | Text classification method and system, electronic equipment and computer readable storage medium |
CN111488459A (en) * | 2020-04-15 | 2020-08-04 | 焦点科技股份有限公司 | Product classification method based on keywords |
CN111985222A (en) * | 2020-08-24 | 2020-11-24 | 平安国际智慧城市科技股份有限公司 | Text keyword recognition method and related equipment |
WO2020244336A1 (en) * | 2019-06-04 | 2020-12-10 | 深圳前海微众银行股份有限公司 | Alarm classification method and device, electronic device, and storage medium |
CN112084308A (en) * | 2020-09-16 | 2020-12-15 | 中国信息通信研究院 | Method, system and storage medium for text type data recognition |
CN112214598A (en) * | 2020-09-27 | 2021-01-12 | 中润普达(十堰)大数据中心有限公司 | Cognitive system based on hair condition |
CN112256865A (en) * | 2019-01-31 | 2021-01-22 | 青岛科技大学 | Chinese text classification method based on classifier |
CN112559748A (en) * | 2020-12-18 | 2021-03-26 | 厦门市法度信息科技有限公司 | Method for classifying stroke record data records, terminal equipment and storage medium |
CN112883159A (en) * | 2021-02-25 | 2021-06-01 | 北京精准沟通传媒科技股份有限公司 | Method, medium, and electronic device for generating hierarchical category label for domain evaluation short text |
CN113869356A (en) * | 2021-08-17 | 2021-12-31 | 杭州华亭科技有限公司 | Method for judging escape tendency of people based on Bayesian classification |
CN114528404A (en) * | 2022-02-18 | 2022-05-24 | 浪潮卓数大数据产业发展有限公司 | Method and device for identifying provincial and urban areas |
CN114564582A (en) * | 2022-02-25 | 2022-05-31 | 苏州浪潮智能科技有限公司 | Short text classification method, device, equipment and storage medium |
CN116956930A (en) * | 2023-09-20 | 2023-10-27 | 北京九栖科技有限责任公司 | Short text information extraction method and system integrating rules and learning models |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8725732B1 (en) * | 2009-03-13 | 2014-05-13 | Google Inc. | Classifying text into hierarchical categories |
CN104850650A (en) * | 2015-05-29 | 2015-08-19 | 清华大学 | Short-text expanding method based on similar-label relation |
WO2016090197A1 (en) * | 2014-12-05 | 2016-06-09 | Lightning Source Inc. | Automated content classification/filtering |
CN106407482A (en) * | 2016-12-01 | 2017-02-15 | 合肥工业大学 | Multi-feature fusion-based online academic report classification method |
CN107066553A (en) * | 2017-03-24 | 2017-08-18 | 北京工业大学 | A kind of short text classification method based on convolutional neural networks and random forest |
-
2018
- 2018-08-21 CN CN201810951636.2A patent/CN109165294B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8725732B1 (en) * | 2009-03-13 | 2014-05-13 | Google Inc. | Classifying text into hierarchical categories |
WO2016090197A1 (en) * | 2014-12-05 | 2016-06-09 | Lightning Source Inc. | Automated content classification/filtering |
CN104850650A (en) * | 2015-05-29 | 2015-08-19 | 清华大学 | Short-text expanding method based on similar-label relation |
CN106407482A (en) * | 2016-12-01 | 2017-02-15 | 合肥工业大学 | Multi-feature fusion-based online academic report classification method |
CN107066553A (en) * | 2017-03-24 | 2017-08-18 | 北京工业大学 | A kind of short text classification method based on convolutional neural networks and random forest |
Non-Patent Citations (1)
Title |
---|
范云杰、刘怀亮: "基于维基百科的中文短文本分类研究", 《现代图书情报技术》 * |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112256865A (en) * | 2019-01-31 | 2021-01-22 | 青岛科技大学 | Chinese text classification method based on classifier |
CN112256865B (en) * | 2019-01-31 | 2023-03-21 | 青岛科技大学 | Chinese text classification method based on classifier |
WO2020244336A1 (en) * | 2019-06-04 | 2020-12-10 | 深圳前海微众银行股份有限公司 | Alarm classification method and device, electronic device, and storage medium |
CN110619363A (en) * | 2019-09-17 | 2019-12-27 | 陕西优百信息技术有限公司 | Classification method for subclass names corresponding to long description of material data |
CN111159414A (en) * | 2020-04-02 | 2020-05-15 | 成都数联铭品科技有限公司 | Text classification method and system, electronic equipment and computer readable storage medium |
CN111488459B (en) * | 2020-04-15 | 2022-07-22 | 焦点科技股份有限公司 | Product classification method based on keywords |
CN111488459A (en) * | 2020-04-15 | 2020-08-04 | 焦点科技股份有限公司 | Product classification method based on keywords |
CN111985222A (en) * | 2020-08-24 | 2020-11-24 | 平安国际智慧城市科技股份有限公司 | Text keyword recognition method and related equipment |
CN111985222B (en) * | 2020-08-24 | 2023-07-18 | 平安国际智慧城市科技股份有限公司 | Text keyword recognition method and related equipment |
CN112084308A (en) * | 2020-09-16 | 2020-12-15 | 中国信息通信研究院 | Method, system and storage medium for text type data recognition |
CN112214598A (en) * | 2020-09-27 | 2021-01-12 | 中润普达(十堰)大数据中心有限公司 | Cognitive system based on hair condition |
CN112214598B (en) * | 2020-09-27 | 2023-01-13 | 吾征智能技术(北京)有限公司 | Cognitive system based on hair condition |
CN112559748A (en) * | 2020-12-18 | 2021-03-26 | 厦门市法度信息科技有限公司 | Method for classifying stroke record data records, terminal equipment and storage medium |
CN112883159A (en) * | 2021-02-25 | 2021-06-01 | 北京精准沟通传媒科技股份有限公司 | Method, medium, and electronic device for generating hierarchical category label for domain evaluation short text |
CN113869356A (en) * | 2021-08-17 | 2021-12-31 | 杭州华亭科技有限公司 | Method for judging escape tendency of people based on Bayesian classification |
CN114528404A (en) * | 2022-02-18 | 2022-05-24 | 浪潮卓数大数据产业发展有限公司 | Method and device for identifying provincial and urban areas |
CN114564582A (en) * | 2022-02-25 | 2022-05-31 | 苏州浪潮智能科技有限公司 | Short text classification method, device, equipment and storage medium |
CN114564582B (en) * | 2022-02-25 | 2024-06-28 | 苏州浪潮智能科技有限公司 | Short text classification method, device, equipment and storage medium |
CN116956930A (en) * | 2023-09-20 | 2023-10-27 | 北京九栖科技有限责任公司 | Short text information extraction method and system integrating rules and learning models |
Also Published As
Publication number | Publication date |
---|---|
CN109165294B (en) | 2021-09-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109165294A (en) | Short text classification method based on Bayesian classification | |
CN108710651B (en) | Automatic classification method for large-scale customer complaint data | |
CN109189901B (en) | Method for automatically discovering new classification and corresponding corpus in intelligent customer service system | |
CN107577688B (en) | Original article influence analysis system based on media information acquisition | |
CN102289522B (en) | Method of intelligently classifying texts | |
CN110825877A (en) | Semantic similarity analysis method based on text clustering | |
CN110781679B (en) | News event keyword mining method based on associated semantic chain network | |
CN110415071B (en) | Automobile competitive product comparison method based on viewpoint mining analysis | |
CN1687924A (en) | Method for producing internet personage information search engine | |
CN109657063A (en) | A kind of processing method and storage medium of magnanimity environment-protection artificial reported event data | |
CN111309864A (en) | User group emotional tendency migration dynamic analysis method for microblog hot topics | |
CN109783623A (en) | The data analysing method of user and customer service dialogue under a kind of real scene | |
CN112149422B (en) | Dynamic enterprise news monitoring method based on natural language | |
CN111782806A (en) | Artificial intelligence algorithm-based similar marketing enterprise retrieval classification method and system | |
CN111522950B (en) | Rapid identification system for unstructured massive text sensitive data | |
CN115952292B (en) | Multi-label classification method, apparatus and computer readable medium | |
CN115017887A (en) | Chinese rumor detection method based on graph convolution | |
CN111651566A (en) | Multi-task small sample learning-based referee document dispute focus extraction method | |
CN111460147A (en) | Title short text classification method based on semantic enhancement | |
CN108399238A (en) | A kind of viewpoint searching system and method for fusing text generalities and network representation | |
CN105677888A (en) | Service preference identification method based on user time fragments | |
CN109871889B (en) | Public psychological assessment method under emergency | |
CN115600602B (en) | Method, system and terminal device for extracting key elements of long text | |
CN107122420A (en) | A kind of tourist hot spot event detecting method and system | |
CN108804524B (en) | Emotion distinguishing and importance dividing method based on hierarchical classification system |
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 | ||
CP02 | Change in the address of a patent holder | ||
CP02 | Change in the address of a patent holder |
Address after: 241000 room 01, 18 / F, iFLYTEK intelligent building, No. 9, Wenjin West Road, Yijiang District, Wuhu City, Anhui Province Patentee after: ANHUI XUNFEI INTELLIGENT TECHNOLOGY Co.,Ltd. Address before: 241000 Floor 9, block A1, Wanjiang Fortune Plaza, Jiujiang District, Wuhu City, Anhui Province Patentee before: ANHUI XUNFEI INTELLIGENT TECHNOLOGY Co.,Ltd. |