Hydrology worker's professional knowledge examining method and system
Technical field
The present invention relates to internet arena, more particularly to hydrology worker's professional knowledge examining method and system regions.
Background technology
On-line examination of the prior art, generally all it is manually to be selected a topic in exam pool, is then set a question, the examination gone out
Topic is all identical, and then establishment officer takes an exam, and manually carries out sentencing volume, whole process cycle length, cost of labor is high, no
Enough intelligence;And, in the case of trans-regional examined, can not ensure that examination question does not leak due to all examination questions all, and by
It is manually to set a question in examination question, the fairness of examination can not be ensured, existing check-up system also has a kind of defect, is exactly current pin
The examination of every profession and trade, the personnel checking-up needs of different stage are variant, targetedly, just can guarantee that the fairness of examination.
Therefore, in the prior art the defects of is existing examination, using the Assessment manually set a question, not enough intelligently,
And examination paper does not have otherness and specific aim, cause to examine inadequate fairness, cost of labor height.
The content of the invention
For above-mentioned technical problem, the present invention provides a kind of hydrology worker professional knowledge examining method and system, by pre-
Set a question if rule carries out smart random, carry out on-line automatic examination, not only save cost of labor, also improve the fairness of examination.
In order to solve the above technical problems, technical scheme provided by the invention is:
In a first aspect, the present invention provides a kind of hydrology worker professional knowledge examining method, including:
Step S1, in examination paper administration page, obtain the examination paper selection instruction in examination paper list of categories, the examination paper classification
Including hydrology boatman technician, the advanced work of hydrology boatman, hydrology boatman's middle rank work, hydrology boatman's primary work, hydrology exploration work skill
Teacher, the hydrology survey advanced work, the hydrology surveys intermediate work and the hydrology surveys primary work;
Step S2, according to the examination paper selection instruction, according to rule set in advance, random generation is examined from knowledge base
Volume, obtains random examination paper, is shown in examination interface, the knowledge base includes the knowledge base of multiple professional domains, described random
Examination paper includes a variety of topic types;
Step S3, according to the random examination paper, examinee is obtained in the given time generation is replied by the random examination paper
Answer, and automatically generate scoring, the answer and the scoring are shown in the examination interface, complete examination.
Hydrology worker professional knowledge examining method provided by the invention, its technical scheme are:In examination paper administration page, obtain
Examination paper list of categories examination paper selection instruction, the examination paper classification include hydrology boatman technician, the advanced work of hydrology boatman,
Hydrology boatman's middle rank work, hydrology boatman's primary work, hydrology exploration work technician, the hydrology surveys advanced work, the hydrology surveys intermediate work and
The hydrology surveys primary work;According to the examination paper selection instruction, according to rule set in advance, random generation is examined from knowledge base
Volume, obtains random examination paper, is shown in examination interface, the knowledge base includes the knowledge base of multiple professional domains, described random
Examination paper includes a variety of topic types;According to the random examination paper, examinee is obtained in the given time and is replied by the random examination paper
The answer of generation, and scoring is automatically generated, the answer and the scoring are shown in the examination interface, complete examination.
Hydrology worker's professional knowledge examining method of the present invention, set a question carrying out smart random by preset rules, carry out
On-line automatic examination, cost of labor is not only saved, also improve the fairness of examination.
Further, in the step S2, according to the examination paper of regular random set in advance generation, random examination paper is obtained,
Specially:
According to the examination paper classification in the examination paper selection instruction, the paper title of examination paper to be generated is obtained;
Obtain the setting instruction of the examination paper to be generated, the difficulty or ease journey for setting instruction to include the examination paper to be generated
Degree sets instruction, examination paper type instruction, the examination paper quantity in each examination paper type, knowledge category and corresponding score value to set instruction;
Instructed according to described set, the selection of examination paper is carried out in the knowledge base, is obtained in the examination paper to be generated
All examination paper, difficulty or ease grade is carried out all examination paper according to corresponding to examination paper type and each examination paper in the knowledge base
Classification storage;
According to all examination paper in the examination paper to be generated, random examination paper is generated.
Further, in addition to:Online updating is carried out to the knowledge base:
New knowledge is obtained by background management system;
Classification is identified to new knowledge, obtains professional domain classification results corresponding to the new knowledge;
According to the professional domain classification results, the new knowledge is increased into corresponding professional domain knowledge base, it is real
Now to the online updating of knowledge base.
Further, in addition to:The knowledge base is extended:
Obtain the knowledge base of extension to be increased;
The knowledge base of the extension to be increased is increased in the knowledge base, each professional domain in the knowledge base
Knowledge base stored respectively.
Further, in addition to:Printing step, it is specially:
In the examination interface, print command is obtained;
According to the print command, the random examination paper and corresponding answer are printed.
Second aspect, the invention provides a kind of hydrology worker professional knowledge appraisal system, including:
Examination paper selection instruction acquisition module, for being selected in examination paper administration page, the examination paper obtained in examination paper list of categories
Instruction, the examination paper classification include hydrology boatman technician, the advanced work of hydrology boatman, hydrology boatman's middle rank work, hydrology boatman primary
Work, hydrology exploration work technician, the hydrology surveys advanced work, the hydrology surveys intermediate work and the hydrology surveys primary work;
Random examination paper generation module, for according to the examination paper selection instruction, according to rule set in advance from knowledge base
In random generation examination paper, obtain random examination paper, be shown in examination interface, the knowledge base includes knowing for multiple professional domains
Know storehouse, the random examination paper includes a variety of topic types;
Module is completed in examination, for according to the random examination paper, obtaining examinee in the given time and being examined at random by described
Volume replies the answer of generation, and automatically generates scoring, and the answer and the scoring are shown in the examination interface, complete to examine
Core.
A kind of hydrology worker professional knowledge appraisal system provided by the invention, its technical scheme are:Referred to by examination paper selection
Acquisition module is made, in examination paper administration page, obtains the examination paper selection instruction in examination paper list of categories, the examination paper classification includes
Hydrology boatman technician, the advanced work of hydrology boatman, hydrology boatman's middle rank work, hydrology boatman's primary work, hydrology exploration work technician, water
Text surveys advanced work, the hydrology surveys intermediate work and the hydrology surveys primary work;By random examination paper generation module, examined according to described
Selection instruction is rolled up, examination paper is generated according to rule set in advance at random from knowledge base, random examination paper is obtained, is shown in examination
Interface, the knowledge base include the knowledge base of multiple professional domains, and the random examination paper includes a variety of topic types;By examining
Core completes module, according to the random examination paper, obtains examinee in the given time and replies answering for generation by the random examination paper
Case, and scoring is automatically generated, the answer and the scoring are shown in the examination interface, complete examination.
Hydrology worker's professional knowledge appraisal system of the present invention, set a question carrying out smart random by preset rules, carry out
On-line automatic examination, cost of labor is not only saved, also improve the fairness of examination.
Further, the random examination paper generation module, is examined specifically for what is generated according to regular random set in advance
Volume, obtains random examination paper:
According to the examination paper classification in the examination paper selection instruction, the paper title of examination paper to be generated is obtained;
Obtain the setting instruction of the examination paper to be generated, the difficulty or ease journey for setting instruction to include the examination paper to be generated
Degree sets instruction, examination paper type instruction, the examination paper quantity in each examination paper type, knowledge category and corresponding score value to set instruction;
Instructed according to described set, the selection of examination paper is carried out in the knowledge base, is obtained in the examination paper to be generated
All examination paper, difficulty or ease grade is carried out all examination paper according to corresponding to examination paper type and each examination paper in the knowledge base
Classification storage;
According to all examination paper in the examination paper to be generated, random examination paper is generated.
Further, in addition to knowledge base update module, it is specifically used for:
New knowledge is obtained by background management system;
Classification is identified to new knowledge, obtains professional domain classification results corresponding to the new knowledge;
According to the professional domain classification results, the new knowledge is increased into corresponding professional domain knowledge base, it is real
Now to the online updating of knowledge base.
Further, in addition to knowledge base expansion module, it is specifically used for:
Obtain the knowledge base of extension to be increased;
The knowledge base of the extension to be increased is increased in the knowledge base, each professional domain in the knowledge base
Knowledge base stored respectively.
Further, in addition to print module, it is specifically used for:
In the examination interface, print command is obtained;
According to the print command, the random examination paper and corresponding answer are printed.
Compared with prior art, beneficial effects of the present invention are:
Hydrology worker's professional knowledge examining method and system based on the present invention, professional knowledge is designed for hydrological industry
Software is practised and checked and rated, utilizes computer technology, data traversal retrieval and random generation, the dynamically new and high technology such as processing, complete record
Enter Occupational Technique Authentication test item bank-water conservancy point storehouse《The hydrology surveys work examination question collection》And Occupational Technique Authentication examination question
Storehouse-water conservancy divides storehouse《The hydrology surveys boatman's examination question collection》, self-defined paper topic quantity and score value can be often inscribed on request, from huge
Database in randomly select examination question generation paper, realize the on-line study of hydrology professional knowledge, automatic scoring, by specify will
Seek random generation examination paper and generate corresponding Key for Reference.
As computer, communication, network are the rapid development of new and high technology and the extensive use of information technology of mark,
In order to improve learning efficiency, establish the comprehensive examination and evaluation system based on the B/S network architectures, realize each level candidate examination
Accuracy and fairness, accuracy is improved, be a set of practical, quick, computer that automaticity is high side of examination online
Method and system, knowledge annual screening is carried out to worker's business, to be supplied to the business evaluating department more accurately result of appraisal.
Brief description of the drawings
, below will be to tool in order to illustrate more clearly of the specific embodiment of the invention or technical scheme of the prior art
The required accompanying drawing used is briefly described in body embodiment or description of the prior art.
Fig. 1 shows a kind of flow chart for hydrology worker professional knowledge examining method that the embodiment of the present invention is provided;
Fig. 2 shows a kind of schematic diagram for hydrology worker professional knowledge appraisal system that the embodiment of the present invention is provided.
Embodiment
The embodiment of technical solution of the present invention is described in detail below in conjunction with accompanying drawing.Following examples are only used
In clearly illustrating technical scheme, therefore example is intended only as, and the guarantor of the present invention can not be limited with this
Protect scope.
Embodiment one
Fig. 1 shows a kind of flow for hydrology worker professional knowledge examining method that first embodiment of the invention is provided
Figure;As shown in figure 1, a kind of hydrology worker professional knowledge examining method that embodiment one provides, including:
Step S1, in examination paper administration page, the examination paper selection instruction in examination paper list of categories is obtained, examination paper classification includes
Hydrology boatman technician, the advanced work of hydrology boatman, hydrology boatman's middle rank work, hydrology boatman's primary work, hydrology exploration work technician, water
Text surveys advanced work, the hydrology surveys intermediate work and the hydrology surveys primary work;
Step S2, according to examination paper selection instruction, examination paper is generated according to rule set in advance at random from knowledge base, obtained
To random examination paper, examination interface is shown in, knowledge base includes the knowledge base of multiple professional domains, and random examination paper includes more
Kind topic type;
Step S3, according to random examination paper, the answer that examinee replies generation by random examination paper is obtained in the given time,
And scoring is automatically generated, answer and scoring are shown in examination interface, complete examination.
Hydrology worker professional knowledge examining method provided by the invention, its technical scheme are:In examination paper administration page, obtain
The examination paper selection instruction in examination paper list of categories is obtained, examination paper classification includes hydrology boatman technician, the advanced work of hydrology boatman, the hydrology
Boatman's middle rank work, hydrology boatman's primary work, hydrology exploration work technician, the hydrology survey advanced work, the intermediate work of hydrology exploration and the hydrology
Survey primary work;According to examination paper selection instruction, according to rule set in advance from knowledge base generation examination paper at random, obtain with
Machine examination paper, examination interface is shown in, knowledge base includes the knowledge base of multiple professional domains, and random examination paper includes a variety of topics
Type;According to random examination paper, examinee is obtained in the given time the answer of generation is replied by random examination paper, and automatically generated and comment
Point, answer and scoring are shown in examination interface, complete examination.
Hydrology worker's professional knowledge examining method of the present invention, set a question carrying out smart random by preset rules, carry out
On-line automatic examination, cost of labor is not only saved, also improve the fairness of examination.
Wherein, various papers are shown in examination management interface, including the paper examined and the paper do not examined.With each examination
The name of volume is referred to as the differentiation of paper.
Preferably, in step S2, according to the examination paper of regular random set in advance generation, random examination paper is obtained, specifically
For:
Examination paper classification in examination paper selection instruction, obtain the paper title of examination paper to be generated;
The setting instruction of examination paper to be generated is obtained, sets instruction to include the complexity of examination paper to be generated and instruction is set, examined
Inscribe type instruction, the examination paper quantity in each examination paper type, knowledge category and corresponding score value and instruction is set;
Instructed according to setting, the selection of examination paper is carried out in knowledge base, obtains all examination paper in examination paper to be generated, institute
Having examination paper, difficulty or ease grade carries out classification storage according to corresponding to examination paper type and each examination paper in knowledge base;
All examination paper in examination paper to be generated, generate random examination paper.
Wherein, examination paper type instruction can be configured according to the concrete condition of examinee, such as hydrology boatman's primary work
It can set a question from hydrology boatman's primary work exam pool, can also set a question from hydrology boatman's middle rank work exam pool, can also be from hydrology ship
Set a question in the advanced work exam pool of work, optional at least two exam pool can also be set a question in three exam pools, the people that sets a question can be from traveling
Row selection.
Wherein, examination paper quantity and corresponding score value can be also configured by the people that sets a question is self-defined, for example total score 100 is divided, root
According to the setting of topic type, topic type includes the topic types such as single choice, multiple choice, simple answer, gap-filling questions and True-False;Corresponding different topic
Type, different fractions is set.
In addition, the people that sets a question can set the complexity of paper, the examination question stored in knowledge base has corresponding difficulty or ease
Grade classification, different grade of difficulty is set, will be selected a topic according to the grade of difficulty of setting, the topic of each type can
With degree-of-difficulty factor, knowledge category and examination question quantity corresponding to setting;By above-mentioned self-defined setting, make the selected topic more intelligent, more
Targetedly.
Preferably, in addition to:Online updating is carried out to knowledge base:
New knowledge is obtained by background management system;
Classification is identified to new knowledge, obtains professional domain classification results corresponding to new knowledge;
According to professional domain classification results, new knowledge is increased into corresponding professional domain knowledge base, realized to knowledge
The online updating in storehouse.
New knowledge can crawl from website, can also obtain, be uploaded to before knowledge base from other channels, first right
New knowledge is classified, and the keyword in specific extractable new knowledge, is classified according to keyword, made in knowledge base
Data obtain unified management, while ensure the data real-time update in knowledge base, are favorably improved the occupation for treating examination personnel
Attainment and stock of knowledge ability.
Preferably, in addition to:Knowledge base is extended:
Obtain the knowledge base of extension to be increased;
The knowledge base of extension to be increased is increased in knowledge base, the knowledge base of each professional domain is carried out in knowledge base
Store respectively.
Hydrology worker's professional knowledge examining method of the present invention is applied to every field, each professional examination, as long as will
Knowledge base does sufficient division, lays in enough knowledge, it is possible to obtains different papers, the personnel of different field are entered
Row examination, the scope of application are more extensive.Therefore specialized management is set in the present invention, for adding the exam pool in different majors field,
Such as driver's specialty, telemetry communication and Automation Specialty, ship-handling specialty etc..
More specifically, the work post that can be correspondingly arranged under each specialty under different work posts, such as hydrology specialty includes water
Text exploration work, hydrology boatman, driver and comprehensive several work posts, each work post correspond to different grades, such as hydrology exploration work
It is divided into the hydrology and surveys primary work, the intermediate work of hydrology exploration, the advanced work of hydrology exploration and hydrology exploration technician;Hydrology boatman is divided into
Hydrology boatman's primary work, hydrology boatman's middle rank work, the advanced work of hydrology boatman and hydrology boatman technician.Set a question personnel or exam pool pipe
Reason personnel can be managed to the work post grade of division, including increase, delete and change.Set a question personnel or item bank management personnel
Chapters and sections can be also managed, different disciplines correspond to different subject chapters and sections, such as corresponding to advanced work subject setting
Chapters and sections are traffic law, and it is regimen alarm that the hydrology, which surveys the chapters and sections that primary work is set, and the chapters and sections that exploration work synthesis is set are water level
Observation.
Preferably, in addition to:Printing step, it is specially:
In examination interface, print command is obtained;
According to print command, random examination paper and corresponding answer are printed.
The personnel of hydrology worker professional knowledge examination directly can carry out random examination paper and corresponding answer in examination interface
Printing, is easy to the preferred study of examination personnel.
Preferably, personnel are examined can also to carry out online exercise by above-mentioned hydrology worker professional knowledge examining method.
Preferably, after the personnel that set a question set a question, online preview can be carried out, and preserve paper.
Preferably, the personnel that set a question can carry out the change of paper after setting a question, it is more convenient to use paper.
Preferably, examination personnel are treated before examination, to fill in personal information, it is convenient to be preserved for total marks of the examination
Record.
Preferably, administrative staff can check the examination situation of paper by backstage examination paper management system, it is known that who
Member does not participate in examination, and which personnel has completed to examine.
Preferably, knowledge base is divided into multiple knowledge bases according to topic type in the present invention, including it is single choice test items storehouse, multinomial
Selection exam pool, judge exam pool, exam pool of filling a vacancy, explanation of nouns exam pool and simple answer exam pool.In the corresponding management field of each exam pool
Face, addition examination question and corresponding answer are can customize, sets the complexity per problem and the knowledge category.The examination gone out
Volume is stored in managing test paper interface in the form of a list, the examination paper that went out before the personnel that set a question can inquire about, essentially according to examining
Roll up classification and examination chapters and sections inquiry, paper corresponding to online browse;Paper can be also directly added at managing test paper interface, is carried out
The formulation of new paper.
Preferably, item bank management personnel can also by log in log query log in examination personnel's account, landing time and
IP address, which is understood based on this treat that examination personnel have logged on and examined, which does not participate in examination also.
Preferably, all functions of carrying out self-defined setting are corresponding all sets authority to be managed, different volume authorities
Grade can perform different operations, make the management more specification of knowledge base.
Second aspect, the invention provides a kind of hydrology worker professional knowledge appraisal system 10, including:
Examination paper selection instruction acquisition module 101, in examination paper administration page, obtaining the examination paper in examination paper list of categories
Selection instruction, examination paper classification include hydrology boatman technician, the advanced work of hydrology boatman, hydrology boatman's middle rank work, hydrology boatman primary
Work, hydrology exploration work technician, the hydrology surveys advanced work, the hydrology surveys intermediate work and the hydrology surveys primary work;
Random examination paper generation module 102, for according to examination paper selection instruction, according to rule set in advance from knowledge base
In random generation examination paper, obtain random examination paper, be shown in examination interface, knowledge base includes the knowledge base of multiple professional domains,
Random examination paper includes a variety of topic types;
Module 103 is completed in examination, for according to random examination paper, obtaining examinee in the given time and being answered by random examination paper
The answer of repetitive generation, and scoring is automatically generated, answer and scoring are shown in examination interface, complete examination.
A kind of hydrology worker professional knowledge appraisal system 10 provided by the invention, its technical scheme are:Selected by examination paper
Instruction acquisition module 101, in examination paper administration page, the examination paper selection instruction in examination paper list of categories is obtained, examination paper classification includes
Hydrology boatman technician, the advanced work of hydrology boatman, hydrology boatman's middle rank work, hydrology boatman's primary work, hydrology exploration work technician, water
Text surveys advanced work, the hydrology surveys intermediate work and the hydrology surveys primary work;By random examination paper generation module 102, according to examination paper
Selection instruction, examination paper is generated according to rule set in advance at random from knowledge base, obtains random examination paper, is shown in examination circle
Face, knowledge base include the knowledge base of multiple professional domains, and random examination paper includes a variety of topic types;Module is completed by examining
103, according to random examination paper, examinee is obtained in the given time the answer of generation is replied by random examination paper, and automatically generated
Scoring, answer and scoring are shown in examination interface, complete examination.
Hydrology worker's professional knowledge appraisal system 10 of the present invention, set a question carrying out smart random by preset rules, enter
The on-line automatic examination of row, not only saves cost of labor, also improves the fairness of examination.
Preferably, random examination paper generation module 102, specifically for the examination paper generated according to regular random set in advance,
Obtain random examination paper:
Examination paper classification in examination paper selection instruction, obtain the paper title of examination paper to be generated;
The setting instruction of examination paper to be generated is obtained, sets instruction to include the complexity of examination paper to be generated and instruction is set, examined
Inscribe type instruction, the examination paper quantity in each examination paper type and corresponding score value and instruction is set;
Instructed according to setting, the selection of examination paper is carried out in knowledge base, obtains all examination paper in examination paper to be generated, institute
Having examination paper, difficulty or ease grade carries out classification storage according to corresponding to examination paper type and each examination paper in knowledge base;
Preferably, in addition to knowledge base update module, it is specifically used for:
New knowledge is obtained by background management system;
Classification is identified to new knowledge, obtains professional domain classification results corresponding to new knowledge;
According to professional domain classification results, new knowledge is increased into corresponding professional domain knowledge base, realized to knowledge
The online updating in storehouse.
All examination paper in examination paper to be generated, generate random examination paper.
Preferably, in addition to knowledge base expansion module, it is specifically used for:
Obtain the knowledge base of extension to be increased;
The knowledge base of extension to be increased is increased in knowledge base, the knowledge base of each professional domain is carried out in knowledge base
Store respectively.
Preferably, in addition to print module, it is specifically used for:
In examination interface, print command is obtained;
According to print command, random examination paper and corresponding answer are printed.
Embodiment two
Based on the hydrology worker's professional knowledge examining method and system in embodiment one, the generation of random examination paper will have
Randomness, fairness, and the examination question in examination paper does not have deviation, is all investigating scope, based on this, following improvement is being done, based on pass
Keyword carries out the positioning of corresponding knowledge base, then generates examination paper at random in specified knowledge base.
It is specific as follows:
The candidate keywords in retrieval request are determined according to the prediction weight of basic keyword in knowledge base, wherein basis
The prediction weight of keyword is determined according to structural information of the basic keyword in the document of knowledge base, in knowledge base;
The theme affiliated in knowledge base according to candidate keywords, determines other expanded keywords;
Retrieved according to candidate keywords and expanded keyword in knowledge base.
Wherein basic keyword and expanded keyword refer to the keyword for confirming to generate random examination paper exam pool scope, lead to
Cross keyword and navigate to corresponding knowledge base, generate examination paper at random.
Preferably, in addition to:
Non-supervisory keyword abstraction method based on figure, keyword abstraction is carried out to the document in knowledge base, obtains knowledge
The basic keyword set in storehouse, and generate the statistical weight of basic keyword in basic keyword set and structure letter in a document
Breath;
Using two sorting algorithms of setting, structural information according to resulting basic keyword and its in a document, obtain
To the prediction weight of basic keyword;
Wherein, statistical weight is the ratio of the quantity of document and total number of documents amount where basic keyword in knowledge base.
Wherein, the candidate keywords in retrieval request are determined according to the prediction weight of basic keyword in knowledge base, are wrapped
Include:
Basic keyword of the search with segmenting matching in retrieval request from basic keyword set, obtain the basis of matching
The prediction weight and statistical weight of keyword;
The statistical weight of basic keyword to being matched is weighted with prediction weight, the matched basis pass of generation
The new weight of keyword;
The basic keyword matched under new weight satisfaction is imposed a condition is as candidate keywords.
Wherein, two classification set is two classification based on supporting vector machine model, two classification based on maximum entropy
Two classification of method or logic-based regression model.
Wherein, structural information of the basic keyword in knowledge base includes position of the basic keyword in knowledge base, base
The part of speech of the part of speech of plinth keyword, the part of speech of previous word and/or the latter word.
Preferably, in addition to:Document in knowledge base is segmented, generated by being segmented in a document in knowledge base
The matrix of weight composition;
The document in knowledge base is trained using topic model, is the theme vector group by segmenting by matrix decomposition
Into the first matrix and the second matrix being made up of the theme vector of document product, wherein, the theme vector of participle is by segmenting
Weight composition in theme, the theme vector of document are made up of the weight of theme in a document.
Wherein, the theme affiliated in knowledge base according to candidate keywords, determines other expanded keywords, including:
The theme vector of query candidate keyword from the first matrix, candidate is determined according to the theme vector that inquiry obtains
The maximum preceding M theme of keyword distribution of weights;
The participle vector of M theme is inquired about from the first matrix, the participle vector obtained according to inquiry determines M theme
In the maximum top n participle of theme distribution weight, the expanded keyword as corresponding theme;
Wherein, M and N is natural number.
Wherein, retrieved according to candidate keywords and expanded keyword in knowledge base, including:
One new theme vector is determined according to the theme vector of candidate keywords and expanded keyword in the first matrix;
The destination document in knowledge base is determined according to the theme vector of new theme vector and document.
Wherein, a new theme is determined according to the theme vector of candidate keywords and expanded keyword in the first matrix
Vector, the destination document in knowledge base is determined according to the theme vector of new theme vector and document, including:
By the expanded keyword corresponding with the theme of the theme vector of candidate keywords corresponding to theme in M theme
Theme vector is weighted, and obtains theme vector collection, wherein weighted factor candidate key according to corresponding to theme in M theme
Weight of the word in the theme obtains;
The theme vector that theme vector is concentrated is normalized after addition, obtains new theme vector;
The theme vector of document in new theme vector and the second matrix is subjected to Similarity Measure, according to similarity
Result of calculation determine destination document in knowledge base.
According to the structural information of basic keyword in knowledge base, the prediction weight of basic keyword is obtained, according to resulting
Prediction weight determine candidate keywords in retrieval request, can so treat each participle in retrieval request with a certain discrimination, extract
To the candidate keywords that can express user view so that retrieval result accuracy rate is higher;According to candidate keywords in document library
In belonging to theme, determine other expanded keywords, examined according to candidate keywords and expanded keyword in document library
Rope, it is achieved thereby that to retrieval of the retrieval request based on semantic level, can accurately, comprehensively extract and represent user's request
Document, while the random paper randomness of generation is good, knowledge covering is accurate, zero deflection.
Embodiment three
Based on the hydrology worker's professional knowledge examining method and system in embodiment one and embodiment two, for knowledge base
Renewal or extension, classification is identified to the new knowledge extracted, what is only identified is accurate, can just assign to corresponding to
In knowledge base, the classification based on this present embodiment to new knowledge is improved, and concrete scheme is as follows, trains a transduction
Grader, new knowledge is classified:
Reception has mark data points, and each has mark data points to have at least one mark, represent the data point be by
Include the training examples of the data point of a classification specified, or the instruction for the data point being excluded from a classification specified
Practice sample;
Receive data untagged point;
Receiving has at least one default cost factor of mark data points and data untagged point;
By iterative calculation, using at least one cost factor, and there are mark data points and data untagged point conduct
Training examples, using maximum entropy-discriminate (MED), a transductive classifier is trained, wherein, for iterating to calculate each time, regulation
Function of the data untagged point cost factor as an expectation mark value, and estimating according to data point group membership's probability
Calculate, adjust a data point markers prior probability;
Using the transductive classifier classification data untagged point of training, have in mark data points and input data point
It is at least one, and the classification of the data point of classification or its derivative data are exported.
Preferably, a Gaussian prior also including the use of decision function parameter, the given training for being included into and being excluded
Sample, marked according to their expectation, by the use of having mark and data untagged as training examples, it is determined that the KL with minimum
The step of decision function of diverging.
Preferably, the multinomial prior also including the use of decision function parameter is distributed, it is determined that the KL divergences with minimum
The step of decision function.
Preferably, the iterative step of one transductive classifier of repetition training, until reaching the convergence of data value.
When the change of the decision function of transductive classifier is dropped to below a default threshold value, reach convergence.
Or when it is determined that the change of expectation mark value drop to below a default threshold value when, reach convergence.
Wherein, function is the absolute value of the expectation mark of a data point.
Wherein, mark data points represent the data with keyword, data untagged point represents the number of no keyword
According to.
The data for having keyword and the data of no keyword are all taken into account, avoids omitting useful information, makes to know
The information for knowing storehouse is imperfect.
As another preferred embodiment, the classification for new knowledge, in addition to following methods:
(1) training dataset of knowledge data is established, extracts different types of feature, and knowledge data is labeled,
Knowledge data represents to include polytype knowledge data;
(2) training dataset is expressed as tensor, obtains the higher-dimension knowledge data classification based on the study of largest interval tensor
Object function, and object function is analyzed and optimized, obtain disaggregated model;
(3) different types of feature is extracted to new knowledge data, according to disaggregated model, new knowledge data marked
Classification.
Wherein, the training dataset of knowledge data is established, extracts different types of feature, and rower is entered to knowledge data
Note, it is specially:
1) knowledge data needed for crawlers download user is write, forms knowledge data set DATA={ D1... ...,
DIN, wherein INIt is the knowledge data number in set;
2) different types of feature, T are extracted to the knowledge data in DATA1..., TN-1, species number that N-1 is characterized;
3) knowledge data in DATA is labeled, positive example is " 1 ", and counter-example is " 0 ";
4) training tensor is establishedWherein I1..., IN-1Mode corresponds to the spy of knowledge data in step 2)
Levy T1..., TN-1, INMode corresponds to knowledge data number.
Wherein, the step of (2), includes:
1) according to training tensor X, the target letter of the higher-dimension knowledge data classification based on the study of largest interval tensor is obtained
Number:
2)
S.t.Un>0,1<=n<=N
Wherein Ω (X) represents the supervision message of training data, Un(1≤n≤N) is obtained matrix after tensor resolution, C
For core tensor, its n rank expansion Matrix C (n) meets following condition:
A) C (n) element is made up of " 0 " or " 1 " entirely;
B) C (n) all rows are mutually orthogonal;
C) for arbitrary n, C (n) is full rank;
2) deployed according to tensor, formula (1) can write:
Wherein, B(n)=Cx1U1x2...xNUN, X (n) is that the n ranks for training tensor X deploy matrix;
Make X(n)=[x1,x2,...,xIn]T,U(n)=[u1,u2,...,uIn]T, by each matrix U in formula (1)iTransposition
And it is divided into IiIndividual independent optimization problem:
3) there will be component during supervision message, i.e. n=N to introduce the grader of largest interval in formula (2) as supervision to believe
Breath, obtains following majorized function:
Wherein, γ be control approximate error weight parameter, λ be control tactics error weight parameter, yiTo be corresponding
Label is marked, α is sorting parameter to be optimized, and L is loss function L (y, t)=max (0,1-yt)2, K is nuclear matrix, its yuan
Plain kij=k (ui, uj), k is kernel function;
4) method for using Conjugate gradient descent, iteratively Optimal Parameters α and matrix component ui (N);
Calculate α gradient first during Optimum Classification parameter alpha:
Wherein I0For IN×INDiagonal matrix, wherein preceding nv(each number of supporting vector) individual element is 1, and remaining is 0.
Then α Hessian matrixes are calculated:
Hα=2 (λ K+KI0K)
During matrix component is optimized, it is first assumed that use inner product core:
k(ui (N),uj (N))=ui (N)Tuj (N)
Calculate ui (N)Gradient:
Then the Hessian matrixes calculated:
Wherein, InsIt is that size is IsUnit matrix, [i ∈ nv] it is an indicator function, and if only if, and i belongs to support
Functional value is 1 during the set of vector, and remaining is 0;
5) for during the mode, i.e. n ≠ N of unsupervised information, adding the constraint of sparse selection, i.e. I in formula (2)1Model
Number:
Wherein, η(n)It is to control the degree of rarefication in mode n;
6) solution formula (4) with the following method
Wherein, uij (n)For ui (n)In element,
B(n)=[b1 T,b2 T,...,bRn T]T
T=bj(BT (n)ui (n)-bj Txi)
7) u tried to achieve according to step 4) and step 6)i, U is pieced together, is iterated, until convergence.Obtain disaggregated model
Parameter { U1..., UN;α}.
Wherein, the knowledge data to be sorted needed for crawlers download user 1) is write, forms knowledge data test set
Close TEST={ Dt1,...,DtINt, wherein INtIt is the knowledge data number to be sorted in set TEST;
2) different types of feature is extracted to the multi-medium data in TEST, it is consistent with the feature extracted during training,
Tt1..., TtN-1, species number that N-1 is characterized;
3) test tensor is establishedWherein I1..., IN-1Mode corresponds to multi-medium data in step 2)
Feature T1..., TN-1, INMode corresponds to knowledge data number to be sorted;
4) according to obtained disaggregated model parameter { U1..., UN;α }, and formula (3), calculate multimedia to be sorted
The y of datai;
5) according to the y obtained in step 4)i, the binarization operation for threshold value with 0.5 is carried out, obtains knowledge to be sorted
The label and classification results of data.
Higher-dimension for knowledge data and structural, expresses knowledge data, and pass through largest interval point using tensor
The method of class device, the knowledge data of higher-dimension is classified.Complete to classify while decomposition analysis is carried out to knowledge data,
The structural information in knowledge data is not only remained, and avoids high dimensional data caused by traditional method by split
" dimension disaster " triggered.
Finally it should be noted that:Various embodiments above is merely illustrative of the technical solution of the present invention, rather than its limitations;
Although the present invention is described in detail with reference to foregoing embodiments, it will be understood by those within the art that:Its
The technical scheme described in foregoing embodiments can still be modified, it is either special to which part or whole technologies
Sign carries out equivalent substitution;And these modifications or replacement, the essence of appropriate technical solution is departed from various embodiments of the present invention
The scope of technical scheme, it all should cover among the claim of the present invention and the scope of specification.