CN111476027A - Big data based anti-smuggling case information extraction method - Google Patents
Big data based anti-smuggling case information extraction method Download PDFInfo
- Publication number
- CN111476027A CN111476027A CN202010263448.8A CN202010263448A CN111476027A CN 111476027 A CN111476027 A CN 111476027A CN 202010263448 A CN202010263448 A CN 202010263448A CN 111476027 A CN111476027 A CN 111476027A
- Authority
- CN
- China
- Prior art keywords
- smuggling
- convolution
- information
- sequence
- expansion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 claims abstract description 18
- 230000000694 effects Effects 0.000 claims abstract description 10
- 238000012795 verification Methods 0.000 claims abstract description 8
- 238000005516 engineering process Methods 0.000 claims abstract description 6
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000003058 natural language processing Methods 0.000 claims abstract description 5
- 239000013598 vector Substances 0.000 claims description 32
- 230000006870 function Effects 0.000 claims description 23
- 238000002372 labelling Methods 0.000 claims description 17
- 238000013527 convolutional neural network Methods 0.000 claims description 16
- 230000010339 dilation Effects 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012546 transfer Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 230000013016 learning Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 239000011800 void material Substances 0.000 claims description 2
- 230000011218 segmentation Effects 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 5
- 239000000463 material Substances 0.000 abstract description 4
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010052804 Drug tolerance Diseases 0.000 description 1
- 230000003679 aging effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000026781 habituation Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Tourism & Hospitality (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a big data-based anti-smuggling case information extraction method, provides an anti-smuggling information extraction model facing big data, and verifies the specific application effect of the model in a real case copy. Firstly, constructing an intelligence element expression model of the anti-smuggling case based on the anti-smuggling case information; then, automatic extraction of various smuggling information elements is realized through a natural language processing technology and a deep learning algorithm model; and finally, the intelligent verification of the smuggling information extraction effect under a big data environment is realized by combining the document chain, the fund chain and the goods chain evidence chain. According to the method, the key element model and the natural language processing technology are used for reference, the information related to the smuggling is used as a data source, the smuggling information extraction model under the big data environment is constructed, the smuggling information service method for automatically extracting the smuggling information is formed, experimental verification analysis is carried out, automatic extraction of the information is achieved, and investment of manpower and material resources is effectively saved.
Description
Technical Field
The invention belongs to the technical field of data processing based on a calculation model, and particularly relates to a big data-based anti-smuggling case information extraction method.
Background
Under the background of big data era, informatization means applied by lawless persons in the process of carrying out smuggling and illegal criminal activities is increasing, and relevant information of transactions is stored in various media and mediums, so that the method has the characteristics of large data volume and being hidden in memories in different fields. The specific data sources mainly comprise customs service data, internal data of public security, other administrative data and internet resources.
The criminal activities of smuggling generally involve the declaration of customs documents, waybills, manifest, bill of lading and sale of equivalent textual information materials, as well as the basic personnel information, call records, money incoming and outgoing and related clue information of the criminal of smuggling. In the face of the case massive complex information of various data types in long time span, customs officers must rapidly and efficiently extract effective information related to the case from related materials and arrange evidence chains of fund flow, cargo flow, document flow and the like of the case in time, so as to achieve the purpose of helping to rapidly comb the case and help to solve the case. Each link of the links puts a hard requirement on the extraction capability of customs smuggling information in the big data era, and the information extraction forms a bottleneck for the traditional work of smuggling the information.
Based on the analysis, a method for extracting the anti-smuggling case information, which is suitable for multi-level wide-view large-scale anti-smuggling case data research, is necessary to be constructed in the background of the big data technology.
Disclosure of Invention
The invention provides a big data oriented anti-smuggling information extraction model aiming at the bottleneck problem of information extraction in the traditional anti-smuggling work of customs, and verifies the specific application effect of the model in the real case copy.
In order to achieve the purpose, the technical scheme adopted by the invention is a big data-based wanted-and-private information extraction method, which comprises the following steps:
s1: constructing an intelligence element expression model of the anti-smuggling case based on the anti-smuggling case information;
s2: the automatic extraction of various smuggling information elements is realized through a natural language processing technology and a deep learning algorithm model;
s3: and the intelligent verification of the smuggling information extraction effect under a big data environment is realized by combining the document chain, the fund chain and the goods chain evidence chain.
The step S1 further includes the following steps:
s11: element expression of anti-smuggling case
And constructing an element expression model for the smuggling of the cases based on the case characteristics and the data sources of the smuggling of the cases.
S12: character vectorization
And carrying out element labeling on the wanted case text according to the composition and the relation of the element expression model to obtain a labeled data set, and carrying out unsupervised training learning on the text of the good words by using a related model tool for generating word vectors on the basis of a labeled training set and unlabeled case text linguistic data.
Preferably, the related model tool for generating word vectors uses word2vec for training to reconstruct the linguistic word text.
The modeling process of the deep learning algorithm model in the step S2 specifically includes the following steps:
s21: dilated convolution
Applying the convolutional neural network layer to the sequence labeling problem, performing convolution operation on sequence vectors, performing affine transformation on an input sequence, and performing affine transformation on a subsequence vector x of the sequence vectortIs defined as the output of
(1) In the formula, r is convolution radius, ⊕ is vector linking operation, the expansion convolution increases expansion width on the filter of common convolution, when there is input sequence vector, the expansion convolution can automatically skip all input data in the middle of expansion width, so that effective input becomes wider, more input data can be obtained, when the expansion convolution is used for sequence marking, subsequence vector x is used for sequence markingtIs defined as the output of
(2) In the formula: σ is an expansion width, when σ is 1, the expansion convolution is the same as the ordinary convolution operation, and when σ is larger than 1, the expansion width enables the receptive field range to be larger, so that the void problem caused by the expansion width can be added into the convolution operation through the translation of the filter;
s22: iterative dilation convolutional neural network
The iterative expansion convolutional neural network is used for preventing the output result from having the risk of overfitting through simple layer number superposition, the iterative expansion convolutional neural network is used for multiple times for the same expansion convolutional block, the output result of the last layer is used as input in each iteration, and the same parameters are repeatedly used in each iteration;
s23: conditional random field
The conditional random field defines a series of binary characteristic functions, including a transfer characteristic function and a state characteristic function, the former considers the dependency relationship between output variables, the latter considers the influence of input characteristics, the weighting summation of all the characteristic functions is carried out to obtain the score of each labeling category, the probability of each labeling category is obtained by applying a normalization factor, the maximum probability is the current labeling category, and the observation sequence X is (X is the current labeling category), the observation sequence X is the observation sequence X1,x2,…xn) Under the condition of X, the predicted sequence Y ═ Y | X of the random field P ═ Y | X of the linear chain member1,y2,…yn) The simplified representation of the conditional probability with the value y is
(3) In the formula: z (x) is a normalization factor; exp is an exponential function; w is akRepresentative characteristic function fk(y, x) corresponding weights; k represents the number of the defined characteristic functions;
s24: iterative expansion convolutional neural network combined with conditional random field
The word vectors and the word vectors obtained through preprocessing are used as the input of a model, the input vectors are input into an iterative expansion convolution network after passing through a common convolution layer, iterative blocks share a group of parameters, finally, the network layer outputs each classified score for each sequence, each score sequence is used as the input of a CRF layer, the CRF layer judges each labeled sequence by utilizing a transfer characteristic function considering the output variable dependency relationship, and further optimizes the classification result obtained through iterative expansion convolution.
In the whole process of step S24, the named entities are first kept from being split into separate words in the preprocessing stage, each entity is used as an independent word vector input model, then an iterative expansion convolutional neural network is selected to extract features with higher robustness, then some connections of the network are discarded randomly by dropout to solve overfitting possibly caused by repeated iteration of the network layer, and finally the result obtained by the network layer is further corrected by the conditional random field.
In step S3, the reliability, effectiveness, timeliness, repeatability, and urgency of the information are comprehensively evaluated and verified in combination with five dimensions of the information evaluation.
Compared with the prior art, the invention has the following beneficial technical effects:
(1) the invention provides a big data-oriented smuggling information extraction model aiming at the bottleneck problem of the custom traditional smuggling information work by taking the difficulty in the customs smuggling work in the big data era as a starting point, and verifies the specific application effect of the model in the real case copy.
(2) The invention realizes automatic extraction of information based on the expression of the anti-smuggling information elements of the space-time frame, effectively saves the input of manpower and material resources, and effectively solves the difficult problem of handling the policeman at the first line of the anti-smuggling information department by applying the positive influence on the anti-smuggling work by a big data analysis method.
(3) The invention uses element models and natural language processing technology for reference, uses the information related to the smuggling cases as data sources, constructs the smuggling information extraction model under the big data environment, forms a smuggling information service method for automatic extraction of the smuggling information, and carries out experimental verification analysis.
Drawings
FIG. 1 is a flow chart of a depth conditional random field model;
FIG. 2 is a model of information element expression;
FIG. 3 is a diagram of a normal convolution and dilation convolution;
FIG. 4 is a depth conditional random field information extraction model.
Detailed Description
The present invention will now be described in further detail with reference to the accompanying drawings.
Different public security data sources have obvious complementary advantages in the aspects of richness of knowledge, expression abstraction, cognition habituation and the like. From the data form, the smuggling data comprises five types of text, voice, image, video, audio and the like. At present, case and event information extraction aiming at structured data is relatively mature and easy to realize. In contrast, the extraction of unstructured text, voice, image, video and audio data requires further research. In order to obtain more context information in case texts, the invention mainly uses text data to construct a depth conditional random field information extraction model based on the driving of smuggling intelligence elements. The modeling process is shown in fig. 1, and specifically includes:
s11: element expression of anti-smuggling case
At present, the national public security information system usually collects, processes and manages information according to five elements of people, affairs, things, organizations and places. The public security service typically needs to answer several types of basic questions: "who is alert involved? When an alert occurs? Where is the alarm? What did the alarm-related actor? Why do the police-related actor do so? What are items related to the alert? What consequences are caused by a case event? Thus, the public security information can be summarized into five basic elements with case as the core, namely time, place, people, articles and events. Wherein, the time and the place are basic conditions for the existence and the evolution of the three elements of people, articles and events, and have attributes, behaviors, states and process characteristics. It should be noted that the character includes an organization and a virtual character. The case and event element relationship can be divided into two types of concept relationship and characteristic relationship. The concept relationship refers to a semantic relationship among concepts of different elements in the same classification system, and includes an equivalence relationship, an upper-level relationship, a lower-level relationship, a same-level relationship, a correlation relationship, and the like. A feature relationship is a relationship between different feature features, such as a temporal relationship, a spatial relationship, an attribute relationship, a state relationship, a process relationship, and the like.
Generally, the intelligence elements of case and event can be divided into three layers, namely a concept layer, an element layer and an element relation layer according to the types, characteristics and logic relation of the elements. Wherein, the element layer can be divided into three sub-layers, including basic characteristics (time, space, attribute, behavior), state characteristics and process characteristics; the element relation layer can be divided into two sub-layers, namely a concept relation and a characteristic relation. According to the hierarchical division of the case elements, the case element semantic units with different levels can be formed. The intelligence element expression model includes an intelligence element submodel and an intelligence element relation submodel, as shown in FIG. 2.
S12: character vectorization
In order to obtain high-quality word vector features, information element labeling is firstly carried out on data information by a basic-level policeman, and the labeled element type is based on an information element expression model provided by the embodiment, so that a labeled data set is obtained. And performing unsupervised training learning on the text of the good words by using a word2vec tool based on the labeled training set and the unlabeled case text corpus.
S21: dilated convolution
The method is characterized in that the expansion convolution is applied to an image segmentation task, the contradiction that information is lost while a receptive field is increased in a pooled image is solved, and extra calculation is not needed, the general convolution structure with a convolution kernel of 3 × 3 and a receptive field of 3 × 3 is shown in fig. 3(a), the expansion convolution with the convolution kernel size of 3 × 3 and an expansion width of 2 is shown in fig. 3(b), the receptive field is increased to 5 × 5, and the expansion convolution increases the receptive field under the condition of no information loss, so that the expansion convolution neural network is applied to element extraction to solve the problem that long sequence dependency is needed in the element extraction.
The convolutional neural network layer is applied to the sequence labeling problem, and the essential is to perform convolution operation on sequence vectors, which is different from two-dimensional image convolution operation, and is equivalent to performing affine transformation on an input sequence. For its subsequence vector xtIs defined as the output of
(1) Where r is the convolution radius and ⊕ is the vector chaining operation, the dilation convolution adds dilation width to the filter of ordinary convolution, when there is an input sequence vector, the dilation convolution will automatically skip all the input data in the middle of dilation width, making the effective input wider and more input data availabletIs defined as the output of
(2) In the formula: σ is the width of the dilation, when σ is 1, the dilation convolution is the same as the ordinary convolution operation, and when σ >1, the dilation width makes the receptive field range larger, and the resulting hole problem can be added to the convolution operation by the translation of the filter. Therefore, the expansion convolution can acquire more context information of the text compared with the common convolution.
S22: iterative dilation convolutional neural network
The iterative dilation convolutional neural network can prevent the output result from having the risk of overfitting through simple layer number superposition. The iterative dilation convolution neural network applies the same dilation convolution block for multiple times, each iteration takes the output result of the last layer as input, and the same parameters are repeatedly used in each iteration, so that the effective input width can be widened, and the model generalization capability can be enhanced.
S23: conditional random field
The conditional random field defines a series of binary feature functions including a transfer feature function and a state feature function, the former being consideredThe dependencies between output variables are taken into account, which takes into account the influence of input characteristics. And weighting and summing all the characteristic functions to obtain the grade of each labeling category, and obtaining the probability of each labeling category by using a normalization factor, wherein the maximum probability is the current labeling category. In the observation sequence X ═ X1,x2,…xn) Under the condition of X, the predicted sequence Y ═ Y | X of the random field P ═ Y | X of the linear chain member1,y2,…yn) The simplified representation of the conditional probability with the value y is
(3) In the formula: z (x) is a normalization factor; exp is an exponential function; w is akRepresentative characteristic function fk(y, x) corresponding weights; k represents the number of defined feature functions.
S24: iterative expansion convolutional neural network combined with conditional random field
The method comprises the steps of using word vectors and word vectors obtained through preprocessing as input of a model, inputting the input vectors into an iterative expansion convolution network after passing through a common convolution layer, wherein the network in fig. 4 is composed of blocks which are iterated for 4 times and comprise 3 expansion convolution layers, and the iterated blocks share a group of parameters.
S3: analysis and verification of smuggling information
The invention combines five dimensions of information evaluation (table 1) to comprehensively evaluate and analyze the reliability, effectiveness, timeliness, repeatability and emergency degree of the information.
TABLE 1 information evaluation analysis matrix
Intelligence assessment feature | A | B | C |
Practicality of use | With strong pertinence | Has a certain potential | Not applicable to |
Authenticity | True and accurate content | Is more reliable | Is unreliable |
Aging property | Advanced early warning | Timely early warning | No early warning and lag |
Repeatability of | Never appeared | Occasionally appear | Always occur |
Degree of urgency | Very severe and severe alarm | Moderate warning | Light warning |
It should be understood that the above description of specific embodiments is not intended to limit the invention, and any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (6)
1. A big data-based anti-smuggling case information extraction method is characterized by comprising the following steps:
s1: constructing an intelligence element expression model of the anti-smuggling case based on the anti-smuggling case information;
s2: the automatic extraction of various smuggling information elements is realized through a natural language processing technology and a deep learning algorithm model;
s3: and the intelligent verification of the smuggling information extraction effect under a big data environment is realized by combining the document chain, the fund chain and the goods chain evidence chain.
2. The big data based extraction method of wanted and private intelligence as claimed in claim 1, wherein step S1 comprises the following steps:
s11: element expression of anti-smuggling case
Constructing an element expression model of the anti-smuggling case based on case characteristics and data sources of the anti-smuggling case;
s12: character vectorization
And carrying out element labeling on the wanted case text according to the composition and the relation of the element expression model to obtain a labeled data set, and carrying out unsupervised training learning on the text of the good words by using a related model tool for generating word vectors on the basis of the labeled data set and the unlabeled case text linguistic data.
3. The big data based anti-privately intelligence extraction method as claimed in claim 2, wherein the relevant model tool to generate word vectors is word2 vec.
4. The big data-based anti-privately information extraction method according to claim 1, wherein the modeling process of the deep learning algorithm model in the step S2 specifically comprises the following steps:
s21: dilated convolution
Applying the convolutional neural network layer to the sequence labeling problem, performing convolution operation on sequence vectors, performing affine transformation on an input sequence, and performing affine transformation on a subsequence vector x of the sequence vectortIs defined as the output of
(1) In the formula: r is the convolution radius;for vector chaining operation, the expansion convolution increases the expansion width on a filter of common convolution, when an input sequence vector exists, the expansion convolution can automatically skip all input data in the middle of the expansion width, so that effective input becomes wider, more input data can be obtained, and when the expansion convolution is used for sequence marking, a subsequence vector x is usedtIs defined as the output of
(2) In the formula: σ is an expansion width, when σ is 1, the expansion convolution is the same as the ordinary convolution operation, and when σ is larger than 1, the expansion width enables the receptive field range to be larger, so that the void problem caused by the expansion width can be added into the convolution operation through the translation of the filter;
s22: iterative dilation convolutional neural network
The iterative expansion convolutional neural network is used for preventing the output result from having the risk of overfitting through simple layer number superposition, the iterative expansion convolutional neural network is used for multiple times for the same expansion convolutional block, the output result of the last layer is used as input in each iteration, and the same parameters are repeatedly used in each iteration;
s23: conditional random field
The conditional random field defines a series of binary characteristic functions, including a transfer characteristic function and a state characteristic function, the former considers the dependency relationship between output variables, the latter considers the influence of input characteristics, the weighting summation of all the characteristic functions is carried out to obtain the score of each labeling category, the probability of each labeling category is obtained by applying a normalization factor, the maximum probability is the current labeling category, and the observation sequence X is (X is the current labeling category), the observation sequence X is the observation sequence X1,x2,…xn) Under the condition of X, the predicted sequence Y ═ Y | X of the random field P ═ Y | X of the linear chain member1,y2,…yn) The simplified representation of the conditional probability with the value y is
(3) In the formula: z (x) is a normalization factor; exp is an exponential function; w is akRepresentative characteristic function fk(y, x) corresponding weights; k represents the number of the defined characteristic functions;
s24: iterative expansion convolutional neural network combined with conditional random field
The word vectors and the word vectors obtained through preprocessing are used as the input of a model, the input vectors are input into an iterative expansion convolution network after passing through a common convolution layer, iterative blocks share a group of parameters, finally, the network layer outputs each classified score for each sequence, each score sequence is used as the input of a CRF layer, the CRF layer judges each labeled sequence by utilizing a transfer characteristic function considering the output variable dependency relationship, and further optimizes the classification result obtained through iterative expansion convolution.
5. The big data-based anti-privately case intelligence extraction method as claimed in claim 4, characterized in that in the whole process of step S24, named entities are firstly kept in a preprocessing stage and are not split by word segmentation, each entity is used as an independent word vector input model, then an iterative expansion convolutional neural network is selected to extract features with higher robustness, then dropouts are used to randomly discard some connections of the network to solve overfitting possibly caused by repeated iteration of the network layer, and finally a conditional random field is used to further correct the result obtained by the network layer.
6. The big data-based anti-smuggling case intelligence extraction method as claimed in claim 1, wherein in the step S3, the anti-smuggling intelligence extraction effect verification is a comprehensive evaluation verification of five dimensions of reliability, effectiveness, timeliness, repeatability and urgency of intelligence evaluation.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010263448.8A CN111476027A (en) | 2020-04-07 | 2020-04-07 | Big data based anti-smuggling case information extraction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010263448.8A CN111476027A (en) | 2020-04-07 | 2020-04-07 | Big data based anti-smuggling case information extraction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111476027A true CN111476027A (en) | 2020-07-31 |
Family
ID=71749907
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010263448.8A Pending CN111476027A (en) | 2020-04-07 | 2020-04-07 | Big data based anti-smuggling case information extraction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111476027A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112100212A (en) * | 2020-09-04 | 2020-12-18 | 中国航天科工集团第二研究院 | Case scenario extraction method based on machine learning and rule matching |
CN112131860A (en) * | 2020-09-18 | 2020-12-25 | 南京森林警察学院 | Acquisition method of smuggling knowledge |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107464061A (en) * | 2017-08-09 | 2017-12-12 | 郑州市公安局 | A kind of synthesis analysis method for supporting public security investigation commander |
CN108763485A (en) * | 2018-05-25 | 2018-11-06 | 南京大学 | A kind of chain of evidence relational model construction method of the judgement document based on text similarity |
-
2020
- 2020-04-07 CN CN202010263448.8A patent/CN111476027A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107464061A (en) * | 2017-08-09 | 2017-12-12 | 郑州市公安局 | A kind of synthesis analysis method for supporting public security investigation commander |
CN108763485A (en) * | 2018-05-25 | 2018-11-06 | 南京大学 | A kind of chain of evidence relational model construction method of the judgement document based on text similarity |
Non-Patent Citations (2)
Title |
---|
曹依依: "基于命名实体识别的医学术语发现及应用", 《中国优秀硕士学位论文全文数据库医药卫生科技辑》 * |
曹依依等: "基于CNN-CRF的中文电子病历命名实体识别研究", 《重庆邮电大学学报(自然科学版)》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112100212A (en) * | 2020-09-04 | 2020-12-18 | 中国航天科工集团第二研究院 | Case scenario extraction method based on machine learning and rule matching |
CN112131860A (en) * | 2020-09-18 | 2020-12-25 | 南京森林警察学院 | Acquisition method of smuggling knowledge |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11763093B2 (en) | Systems and methods for a privacy preserving text representation learning framework | |
Hu et al. | Review sentiment analysis based on deep learning | |
CN111476027A (en) | Big data based anti-smuggling case information extraction method | |
Mukherjee | Graph-based data mining on social networks | |
Mahajan et al. | Correlating crime and social media: using semantic sentiment analysis | |
Li et al. | Graph contrastive learning with feature augmentation for rumor detection | |
Cebi et al. | Deep learning based security management of information systems: A comparative study | |
dos Santos et al. | A two-stage regularization framework for heterogeneous event networks | |
Kotteti et al. | Rumor detection on time-series of tweets via deep learning | |
Surekha et al. | Digital misinformation and fake news detection using WoT integration with Asian social networks fusion based feature extraction with text and image classification by machine learning architectures | |
Xu et al. | Dual generative network with discriminative information for generalized zero-shot learning | |
Bhatnagar et al. | A sentiment analysis based approach for customer segmentation | |
Shiri et al. | Toward the automated construction of probabilistic knowledge graphs for the maritime domain | |
Torgheh et al. | A Novel Method for Detecting Fake news: Deep Learning Based on Propagation Path Concept | |
CN111476483A (en) | Big data based anti-privatization case information correlation method | |
Dhal et al. | A fine-tuning deep learning with multi-objective-based feature selection approach for the classification of text | |
Nahilia et al. | A new corpus-based convolutional neural network for big data text analytics | |
Faccia et al. | NLP And IR Applications For Financial Reporting And Non-Financial Disclosure. Framework Implementation And Roadmap For Feasible Integration With The Accounting Process | |
Mou et al. | A two stage adaptation framework for frame detection via prompt learning | |
Zhang et al. | Research on knowledge graph for quantification of relationship between enterprises and recognition of potential risks | |
Yandrapati et al. | Classifying informative tweets using feature enhanced pre-trained language model | |
Nguyen et al. | Supervised learning models for social bot detection: Literature review and benchmark | |
Dwivedi et al. | State of the Art Machine Learning Techniques for Detecting Fake News | |
Zhang et al. | Target-dependent event detection: A new task to event extraction from news | |
Shanthakumar et al. | Fake Profile Identification in Online Social Networks Using Machine Learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200731 |