CN107169043A - A kind of knowledge point extraction method and system based on model answer - Google Patents
A kind of knowledge point extraction method and system based on model answer Download PDFInfo
- Publication number
- CN107169043A CN107169043A CN201710272107.5A CN201710272107A CN107169043A CN 107169043 A CN107169043 A CN 107169043A CN 201710272107 A CN201710272107 A CN 201710272107A CN 107169043 A CN107169043 A CN 107169043A
- Authority
- CN
- China
- Prior art keywords
- knowledge
- rule
- model answer
- base
- answer
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2455—Query execution
- G06F16/24564—Applying rules; Deductive queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
Abstract
The invention discloses a kind of knowledge point extraction method based on model answer and system, method includes:Natural language understanding is carried out to topic and corresponding model answer, as a result as source Knowledge Set, the source Knowledge Set is put into an inference machine, and a knowledge base is obtained using the training of drools regulation engines according to rule base, each answer step in the model answer is compared successively according to the knowledge base, extracts and marks knowledge point.Knowledge point extraction method proposed by the invention carries out automatic marking using production rule engine, and using computer by unified with nature language understanding, automated reasoning technology to knowledge point.
Description
Technical field
The present invention relates to automated reasoning technology, Knowledge Extraction field, more particularly to a kind of knowledge based on model answer
Point extraction method and system.
Background technology
Knowledge Extraction refer to recognize, find and extract from digital resource concept, type, the fact and its dependency relation,
Constraint rule, and the step of progress problem solving, the process of rule.(reference《The main technique methods solution that current knowledge is extracted
Analysis》, Zhang Zhixiong etc., modem long jump skill intelligence technology, the 8th phase in 2008) at present, most knowledge point, which is extracted, concentrates on design certainly
Dynamic or automanual algorithm, extracts structured message, i.e., from existing (natural language from unstructured and semi-structured text
Speech) knowledge is extracted in document.Such as, the Liu Xiaojuan of Beijing Normal University is in research and utilization Knowledge Extraction technology, to digital book
The digital information resources of magnanimity carry out a series of processing in shop so that user is easier to inquire about and understood, machine is easier to automatically
Processing.Again such as, GATE, KIM, ArtEquAKT etc. external Knowledge Extraction system is mainly used in the text in digital library
Shelves mark and processing, bioinformatics, for semantic web generate metadata, improve index, retrieval, classification and filtering apply effect
Rate etc..It can be seen that, it is either domestic or external, required for being only limited to extract people from text to the application of " Knowledge Extraction "
Information, although have a natural language understanding correlation technique, but without inference mechanism, it appears single.
The content of the invention
The technical problem to be solved in the present invention is how to provide the model answer with knowledge point mark to teacher to understand
Student automatically extracts to the grasp situation of knowledge point while student can be improved to the study of topic answer, the knowledge point of understanding
Method.
Above-mentioned technical problem is solved, the invention provides a kind of knowledge point extraction method based on model answer, bag
Include following steps:
Natural language understanding is carried out to topic and corresponding model answer, as a result as source Knowledge Set,
The source Knowledge Set is put into an inference machine, and one is obtained using the training of drools regulation engines according to rule base
Knowledge base,
Each answer step in the model answer is compared successively according to the knowledge base, extracts and marks
Knowledge point.
Further, natural language understanding is carried out to topic and corresponding model answer to specifically include:
2-1) entity is marked, by topic and corresponding model answer to text in entity predefined symbol mark
Note, as the template of unified textual form,
2-2) template matches, same XML file is constituted to the template according to same type.
Further, the steps is also included after the source Knowledge Set is put into an inference machine:
Entity and two concepts of relation 3-1) are defined,
3-2) based on including known true and rule production rule engine,
3-3) pass through the knowledge base of production rule engine training one.
Further, the inference rule in the inference machine includes:
4-1) the given known fact is inserted into factbase;
4-2) Land use models matching is matched to the known fact in the rule and factbase in rule base;
If 4-3) multiple rules meet condition and be active simultaneously, there is conflicting rule, simultaneously will
The strictly all rules of conflict is put into conflict set;
4-4) it will be put into by setting order in the conflict set to the rule for handling state of activation and handle conflict;
The rule in the conflict set 4-5) is performed using enforcement engine, above step 4-2 is repeated) to 4-4), until institute
State the rule that conflict set is not on state of activation.
Further, the side being compared successively to each answer step in the model answer according to the knowledge base
Method is as follows:
The text of the model answer after natural language understanding is handled in the Knowledge Set of source is as in knowledge and knowledge base
Content make comparisons;
If the text of the model answer is the known conditions in stem, do not deal with;
If the text of the model answer is by known conditions is derived, each relation institute is searched in knowledge base right
The condition set and rule name answered;
If the matching criteria found in model answer derivation in condition and knowledge base used, by knowledge
Corresponding rule name is extracted in storehouse is labeled as knowledge point automatically.
Further, if the text of the model answer exists as knowledge, father's node of the knowledge is searched, by institute
The condition set that the knowledge is derived in knowledge base is stated, while obtaining the knowledge point for deriving the knowledge rule and being marked, such as
Really the knowledge is not present, then the step is not marked.
Further, if the text of the model answer exists as knowledge, and to same knowledge by different condition sets
Produce, then each knowledge node there are two attributes in the knowledge base, and one is the condition set i.e. father of the knowledge node
Node;Another is that rule name infers knowledge point corresponding to the knowledge, then, choose in the knowledge base with current answer
Text in the of equal value condition set of previous step made inferences as father's node of current knowledge, obtain with the condition set with group
Rule name is marked as knowledge point.
A kind of knowledge point automatic extracting system based on model answer is additionally provided based on the invention described above, including:Knowledge
Collecting unit, knowledge processing element and knowledge point extraction unit,
The knowledge acquisition unit, to carry out natural language understanding, as a result conduct to topic and corresponding model answer
Source Knowledge Set,
The knowledge processing element, the source Knowledge Set is put into an inference machine, and is used according to rule base
The training of drools regulation engines obtains a knowledge base,
The knowledge point extraction unit, to be walked successively to each answer in the model answer according to the knowledge base
Suddenly it is compared, extracts and mark knowledge point.
Further, the knowledge processing element is also used to, and the knowledge base is increased into it by setting rule set constraint
Knowledge number, no longer exists new knowledge generation until being triggered there is no new rule and then terminates reasoning process.
Further, the knowledge point extraction unit, also the process natural language understanding in the Knowledge Set of source to be handled
The text of model answer afterwards is made comparisons as knowledge with the content in knowledge base;
If the text of the model answer is the known conditions in stem, do not deal with;
If the text of the model answer is by known conditions is derived, each relation institute is searched in knowledge base right
The condition set and rule name answered;
If the matching criteria found in model answer derivation in condition and knowledge base used, by knowledge
Corresponding rule name is extracted in storehouse is labeled as knowledge point automatically.
Beneficial effects of the present invention:
Knowledge point extraction method proposed by the invention is adopted by unified with nature language understanding, automated reasoning technology
Production rule engine is used, and automatic marking is carried out to knowledge point using computer., can be in exam pool further for teacher
Handmarking knowledge point is replaced in construction, so as to reduce human resources input;The knowledge point precision extracted by computer is high,
Uniformity is good, no uncertainty.Further for student, it is possible to use knowledge point carries out topic recommendation, and student is in the topic that does wrong
In the case of purpose, by knowledge point it is known that oneself which knowledge point there is not grasp, more preferable results of learning are played.
Brief description of the drawings
Fig. 1 is the method flow schematic diagram in one embodiment of the invention;
Fig. 2 is the method processing schematic diagram of the natural language understanding in Fig. 1;
Fig. 3 is the implementation process schematic diagram of inference machine in Fig. 1;
Fig. 4 is the regular implementation schematic diagram in Fig. 3;
Fig. 5 is to be compared schematic diagram to each answer step in the model answer in Fig. 1;
Fig. 6 is the system structure diagram in one embodiment of the invention;
Fig. 7 is the knowledge base formation figure of the present invention;
Fig. 8 is the Analysis of Knowledge Bases Reasoning schematic flow sheet of the present invention;
Fig. 9 is the extraction method schematic diagram of the present invention.
Embodiment
The principle of the disclosure is described referring now to some example embodiments.It is appreciated that these embodiments are merely for saying
It is bright and help it will be understood by those skilled in the art that with the purpose of the embodiment disclosure and describe, rather than advise model of this disclosure
Any limitation enclosed.Content of this disclosure described here can be implemented in the various modes outside mode described below.
As described herein, term " comprising " and its various variants are construed as open-ended term, it means that " bag
Include but be not limited to ".Term "based" is construed as " being based at least partially on ".Term " one embodiment " it is understood that
For " at least one embodiment ".Term " another embodiment " is construed as " at least one other embodiment ".
It will be appreciated by those skilled in the art that name Entity recognition is that the solution of evaluation and test is extracted according to U.S.'s NIST automated contents
Release, the reference of entitative concept in the text there are three kinds of forms:Name property index, nominal index and pronoun index.In reality
In, name Entity recognition is mainly used in the identification that all kinds such as name, place name and organization name entity.Based on bar
Identification process is regarded as a sequence labelling problem by the name Entity recognition of part random field (CRF), and given text is entered first
Row word segmentation processing, basic mathematical element (geometric element title, function expression, equation etc.) is identified.Life based on CRF
Name entity recognition method belongs to the learning method for having supervision, accordingly, it would be desirable to using the large-scale corpus marked to CRF models
Parameter is trained.
It will be appreciated by those skilled in the art that the rule base includes but is not limited to, the inter-entity in art of mathematics is set
Establish rules then, including but not limited to logic rules, the Numeral Rules, plane relation rule, spatial relation rule etc..
Fig. 1 is that the method in the method flow schematic diagram in one embodiment of the invention, the present embodiment includes the steps:
Step S100 carries out natural language understanding to topic and corresponding model answer, as a result as source Knowledge Set, and step S101 is by institute
The source Knowledge Set of stating is put into an inference machine, and obtains a knowledge base, step using the training of drools regulation engines according to rule base
S102 is compared to each answer step in the model answer successively according to the knowledge base, is extracted and is marked knowledge
Point.Compared with the expert system of existing artificial intelligence, the extraction method of the knowledge point proposed in the present embodiment is added
Natural language understanding, in terms of reasoning, the fact that Analysis of Knowledge Bases Reasoning engine not only relies only on stiff in knowledge base and rule, and
The existing fact can be combined with new knowledge, constantly study produces new knowledge, wider using scope.Meanwhile, this method is with automatically
Extract knowledge point and instead of handmarking, largely alleviate the workload of teacher, and facilitate teacher to check Students ' Learning
Situation;Student also is understood that itself grasp situation to knowledge point.
The method processing schematic diagram that Fig. 2 is the natural language understanding in Fig. 1 is refer to, as preferred in the present embodiment,
Natural language understanding is carried out to topic and corresponding model answer to specifically include:
Step S200 entities are marked, by topic and corresponding model answer to text in entity predefined symbol
Mark, as the template of unified textual form,
Step S201 template matches, same XML file is constituted to the template according to same type.
Specifically, the entity mark in the step S200:The entity in text with the symbol mark defined in advance
Note comes out, and such as marks straight line with line, marks triangle with Triangle, then straight line AB means that into AB&&line, triangle
Shape ABC is expressed as ABC&&Triangle, is used with a kind of unified textual form for template matches.
Specifically, the template matches in the step S201:A type of template constitutes an xml document, such as
The template of function is concentrated in the xml document of function, and the template of analytic geometry is gathered in the template of analytic geometry, with such
Push away.A kind of form that the text recognized from outside obtains with template to match by entity mark, such text acceptance of the bid
In the data that the entity of note will be extracted in Json forms, used for Analysis of Knowledge Bases Reasoning.
The implementation process schematic diagram that Fig. 3 is inference machine in Fig. 1 is refer to, as preferred in the present embodiment, by the source
Knowledge Set also includes the steps after being put into an inference machine:
Step S300 defines entity and two concepts of relation, reality defined in the method model proposed in the present embodiment
Body and two concepts of relation, entity refer to the mathematical concepts such as point, line, surface, expression formula, equation;Relation is referred between entity
Relation, such as vertical relation, parallel relation, similarity relation etc..
Step S301 based on including known true and rule production rule engine,
Step S302 is by the knowledge base of production rule engine training one, and Analysis of Knowledge Bases Reasoning is one in the present embodiment
Individual huge production rule engine, including known true and rule.Wherein it is a known fact that by natural language understanding stem
Get, rule comes from concept, theorem, axiom, formula, solution approach in mathematics etc..
Refer to Fig. 4 is the regular implementation schematic diagram in Fig. 3, is used as preferred in the present embodiment, the inference machine
In inference rule include:
The given known fact is inserted into factbase by step S400;
The matching of step S401 Land use models is matched to the known fact in the rule and factbase in rule base;
If the multiple rules of step S402 meet condition and be active simultaneously, there is conflicting rule, together
When by the strictly all rules of conflict be put into conflict set in;
Step S403 will be put into the conflict set by setting order to the rule for handling state of activation and handle conflict;
Step S404 performs the rule in the conflict set using enforcement engine, repeats above step S402 to S404, directly
The rule of state of activation is not on to the conflict set.
Above-mentioned implementation process, specifically refers to the Analysis of Knowledge Bases Reasoning schematic flow sheet that Fig. 8 is the present invention, and above-mentioned knowledge base is same
When be also a production inference engine, reasoning detailed process is as follows:
The given known fact is inserted into factbase;
Land use models matching is matched to the facts in the regular rules and factbase in rule base;
If multiple rules are satisfied condition and be active simultaneously, that is, there is conflicting rule, by the institute of conflict
It is regular to be put into conflict set;
The rule for handling state of activation is put into conflict set in certain sequence, processing conflict;
The rule in conflict set is performed using enforcement engine, above step is repeated, until conflict set is not on activating shape
The rule of state.
By the constraint of rule set, knowledge base can be expanded towards desired method, finally make knowledge number more next
It is more, be triggered until no longer regular, no longer exist new knowledge and produce, whole reasoning process terminates, just obtained behind
The knowledge base marked for knowledge point.
Refer to Fig. 5 is to be compared schematic diagram to each answer step in the model answer in Fig. 1, according to
The method that the knowledge base is compared to each answer step in the model answer successively is as follows:
The text of the model answer after natural language understanding is handled in the Knowledge Set of step S500 sources as knowledge with
Content in knowledge base is made comparisons;
Whether step S501 is known conditions in stem
If the text of the step S502 model answers is the known conditions in stem, do not deal with;
Whether step S503 is by known conditions derivation
If the text of the step S504 model answers is by known conditions is derived, searched in knowledge base each
Condition set and rule name corresponding to relation;
Whether step S505 conditions match with the condition found in knowledge base
If the matching criteria found in step S506 model answer derivations in condition and knowledge base used,
Then corresponding rule name in knowledge base is extracted and is labeled as knowledge point automatically.
The extraction method schematic diagram that Fig. 9 is the present invention is refer to, if the text of the model answer is deposited as knowledge
Father's node of the knowledge is then being searched, the condition set of the knowledge will be being derived in the knowledge base, this is derived while obtaining
The knowledge point of knowledge rule is simultaneously marked, if the knowledge is not present, the step is not marked.
If the text of the model answer exists as knowledge, and same knowledge is produced by different condition sets, then existed
Each knowledge node has two attributes in the knowledge base, one be the i.e. knowledge node of condition set father's node;It is another
Individual is that rule name infers knowledge point corresponding to the knowledge, then, choose in the knowledge base with the text of current answer
Previous step condition set of equal value makes inferences as father's node of current knowledge, obtains the regular masterpiece with group with the condition set
It is marked for knowledge point.
As shown in figure 9, whether criterion answer is final step, if it is, ending standard answer and standard knowledge storehouse
Comparison work;If it is not, then handling the model answer text after natural language processing, detailed process is as follows:
Content to current procedures searches knowledge of equal value in knowledge base, if the knowledge is present, searches the knowledge
The condition set of the knowledge is derived in father's node, i.e. knowledge base, derives that the knowledge point of the knowledge rule is gone forward side by side while obtaining
Line flag.If the knowledge is not present, the step is not marked.
It might be possible that same knowledge may be produced by different condition sets, therefore each in knowledge base
Knowledge node has two attributes, and one is condition set, i.e. father's node of the knowledge node;One is rule name, that is, is inferred
Knowledge point corresponding to the knowledge.Each knowledge node may correspond to the combination of multiple condition sets and rule name.For such a feelings
Condition, will also test in the searching work to model answer current procedures to the previous step of current procedures, choose knowledge base
In the condition set of equal value with current answer previous step made inferences as father's node of current knowledge, obtain same with this condition set
The rule name of group is marked as knowledge point.
Fig. 6 is a kind of based on model answer in the system structure diagram in one embodiment of the invention, the present embodiment
Knowledge point automatic extracting system, including:Knowledge acquisition unit 1, knowledge processing element 2 and knowledge point extraction unit 3, it is described to know
Know collecting unit 1, it is described as a result as source Knowledge Set to carry out natural language understanding to topic and corresponding model answer
Knowledge processing element 2, drools regulation engines are used the source Knowledge Set is put into an inference machine, and according to rule base
Training obtains a knowledge base, the knowledge point extraction unit 3, to according to the knowledge base successively in the model answer
Each answer step is compared, and is extracted and is marked knowledge point.
Preferably, the knowledge processing element 2 is also used to, and by the knowledge base, by setting rule set constraint increase, it is known
Know number, no longer exist new knowledge generation until being triggered there is no new rule and then terminate reasoning process.
Preferably, the knowledge point extraction unit 3, also to by the Knowledge Set of source after natural language understanding is handled
The text of model answer made comparisons as knowledge with the content in knowledge base;
If the text of the model answer is the known conditions in stem, do not deal with;
If the text of the model answer is by known conditions is derived, each relation institute is searched in knowledge base right
The condition set and rule name answered;
If the matching criteria found in model answer derivation in condition and knowledge base used, by knowledge
Corresponding rule name is extracted in storehouse is labeled as knowledge point automatically.
The knowledge point based on model answer that refer in the knowledge base formation figure that Fig. 7 is the present invention, the present embodiment is automatic
Extraction is that model answer is compared with existing standard knowledge base, and knowledge base is got by the natural language understanding to stem,
Natural language understanding is carried out to topic and corresponding model answer, as a result as described source Knowledge Set, by the source Knowledge Set
It is put into an inference machine, and the knowledge base described in obtaining is trained using drools regulation engines according to rule base, is known according to described
Know storehouse to be successively compared each answer step in the model answer, extract and mark knowledge point.Detailed process is such as
Under:The Latex texts of stem pass through natural language processing (NLP), be understood as a series of relation, these relations are known as source
Know collection input inference machine, using existing rule base, production inference is carried out based on drools rule-based reasonings engine, constantly by source
Knowledge Set produces new knowledge, while reasoning can be continued as source Knowledge Set again by inferring the new knowledge come.So repeatedly, until not
There is new knowledge generation again, then reasoning terminates.A knowledge network is finally given, each node of network is a knowledge, this
Knowledge network is referred to as knowledge base.
It should be appreciated that each several part of the present invention can be realized with hardware, software, firmware or combinations thereof.Above-mentioned
In embodiment, the software that multiple steps or method can in memory and by suitable instruction execution system be performed with storage
Or firmware is realized.If, and in another embodiment, can be with well known in the art for example, realized with hardware
Any one of row technology or their combination are realized:With the logic gates for realizing logic function to data-signal
Discrete logic, the application specific integrated circuit with suitable combinational logic gate circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means to combine specific features, structure, material or the spy that the embodiment or example are described
Point is contained at least one embodiment of the present invention or example.In this manual, to the schematic representation of above-mentioned term not
Necessarily refer to identical embodiment or example.Moreover, specific features, structure, material or the feature of description can be any
One or more embodiments or example in combine in an appropriate manner.
In general, the various embodiments of the disclosure can be with hardware or special circuit, software, logic or its any combination
Implement.Some aspects can be implemented with hardware, and some other aspect can be with firmware or software implementation, and the firmware or software can
With by controller, microprocessor or other computing devices.Although the various aspects of the disclosure be shown and described as block diagram,
Flow chart is represented using some other drawing, but it is understood that frame described herein, equipment, system, techniques or methods can
With in a non limiting manner with hardware, software, firmware, special circuit or logic, common hardware or controller or other calculating
Equipment or some of combination are implemented.
In addition, although operation is described with particular order, but this is understood not to require this generic operation with shown suitable
Sequence is performed or performed with generic sequence, or requires that all shown operations are performed to realize expected result.In some feelings
Under shape, multitask or parallel processing can be favourable.Similarly, although the details of some specific implementations is superincumbent to beg for
By comprising but these are not necessarily to be construed as any limitation of scope of this disclosure, but the description of feature is only pin in
To specific embodiment.Some features described in some embodiments of separation can also in combination be held in single embodiment
OK.Mutually oppose, the various features described in single embodiment can also be implemented separately or to appoint in various embodiments
The mode of what suitable sub-portfolio is implemented.
Claims (10)
1. a kind of knowledge point extraction method based on model answer, it is characterised in that comprise the following steps:
Natural language understanding is carried out to topic and corresponding model answer, as a result as source Knowledge Set,
The source Knowledge Set is put into an inference machine, and a knowledge is obtained using the training of drools regulation engines according to rule base
Storehouse,
Each answer step in the model answer is compared successively according to the knowledge base, extracts and marks knowledge
Point.
2. knowledge point extraction method according to claim 1, it is characterised in that to topic and corresponding model answer
Natural language understanding is carried out to specifically include:
2-1) entity is marked, by topic and corresponding model answer to text in entity predefined sign flag, make
To unify the template of textual form,
2-2) template matches, same XML file is constituted to the template according to same type.
3. knowledge point extraction method according to claim 1, it is characterised in that the source Knowledge Set is put into one and pushed away
Also include the steps after in reason machine:
Entity and two concepts of relation 3-1) are defined,
3-2) based on including known true and rule production rule engine,
3-3) pass through the knowledge base of production rule engine training one.
4. knowledge point extraction method according to claim 3, it is characterised in that the inference rule in the inference machine
Including:
4-1) the given known fact is inserted into factbase;
4-2) Land use models matching is matched to the known fact in the rule and factbase in rule base;
If 4-3) multiple rules meet condition and be active simultaneously, there is conflicting rule, while will conflict
Strictly all rules be put into conflict set in;
4-4) it will be put into by setting order in the conflict set to the rule for handling state of activation and handle conflict;
The rule in the conflict set 4-5) is performed using enforcement engine, above step 4-2 is repeated) to 4-4), until the punching
It is prominent to collect the rule for being not on state of activation.
5. knowledge point extraction method according to claim 1, it is characterised in that according to the knowledge base successively to institute
State the method that each answer step in model answer is compared as follows:
The text of the model answer after natural language understanding is handled in the Knowledge Set of source is as in knowledge and knowledge base
Appearance is made comparisons;
If the text of the model answer is the known conditions in stem, do not deal with;
If the text of the model answer is by known conditions is derived, searched in knowledge base corresponding to each relation
Condition set and rule name;
If the matching criteria found in model answer derivation in condition and knowledge base used, by knowledge base
Corresponding rule name is extracted is labeled as knowledge point automatically.
6. knowledge point extraction method according to claim 5, it is characterised in that if the text of the model answer is made
Exist for knowledge, then search father's node of the knowledge, the condition set of the knowledge will be derived in the knowledge base, obtain simultaneously
Derive the knowledge point of the knowledge rule and be marked, if the knowledge is not present, the step is not marked.
7. knowledge point extraction method according to claim 6, it is characterised in that if the text of the model answer is made
Exist for knowledge, and same knowledge is produced by different condition sets, then each knowledge node has two in the knowledge base
Individual attribute, one be the i.e. knowledge node of condition set father's node;Another is that rule name is inferred corresponding to the knowledge
Knowledge point, then, choose condition set of equal value with previous step in the text of current answer in the knowledge base and be used as current knowledge
Father's node make inferences, obtain with the condition set with organize rule name be marked as knowledge point.
8. a kind of knowledge point automatic extracting system based on model answer, it is characterised in that including:Knowledge acquisition unit, knowledge
Processing unit and knowledge point extraction unit,
The knowledge acquisition unit, to carry out natural language understanding to topic and corresponding model answer, as a result knows as source
Know collection,
The knowledge processing element, the source Knowledge Set is put into an inference machine, and uses drools according to rule base
Regulation engine training obtains a knowledge base,
The knowledge point extraction unit, to be entered successively to each answer step in the model answer according to the knowledge base
Row compares, and extracts and marks knowledge point.
9. knowledge point automatic extracting system according to claim 8, it is characterised in that the knowledge processing element is also used
So that the knowledge base is increased into its knowledge number by setting rule set constraint, is triggered i.e. not until there is no new rule
There is new knowledge to produce again and then terminate reasoning process.
10. knowledge point automatic extracting system according to claim 8, it is characterised in that the knowledge point extraction unit, also
To using the text of the model answer after natural language understanding is handled in the Knowledge Set of source as in knowledge and knowledge base
Content is made comparisons;
If the text of the model answer is the known conditions in stem, do not deal with;
If the text of the model answer is by known conditions is derived, searched in knowledge base corresponding to each relation
Condition set and rule name;
If the matching criteria found in model answer derivation in condition and knowledge base used, by knowledge base
Corresponding rule name is extracted is labeled as knowledge point automatically.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710272107.5A CN107169043A (en) | 2017-04-24 | 2017-04-24 | A kind of knowledge point extraction method and system based on model answer |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710272107.5A CN107169043A (en) | 2017-04-24 | 2017-04-24 | A kind of knowledge point extraction method and system based on model answer |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107169043A true CN107169043A (en) | 2017-09-15 |
Family
ID=59812233
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710272107.5A Pending CN107169043A (en) | 2017-04-24 | 2017-04-24 | A kind of knowledge point extraction method and system based on model answer |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107169043A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107992482A (en) * | 2017-12-26 | 2018-05-04 | 科大讯飞股份有限公司 | Mathematics subjective item answers the stipulations method and system of step |
CN108109453A (en) * | 2017-12-29 | 2018-06-01 | 广州大学 | Work correction device |
CN108182177A (en) * | 2018-01-24 | 2018-06-19 | 谢德刚 | A kind of mathematics knowledge-ID automation mask method and device |
CN108255812A (en) * | 2018-01-16 | 2018-07-06 | 西南大学 | Calculus aptitude tests topic intelligent generation method based on semantic tagger |
CN109783775A (en) * | 2019-01-18 | 2019-05-21 | 广东小天才科技有限公司 | A kind of method and system for the content marking user's corpus |
CN109902160A (en) * | 2019-01-30 | 2019-06-18 | 华中师范大学 | The method and system that circuit topic is answered automatically |
CN109918514A (en) * | 2019-03-21 | 2019-06-21 | 江西风向标教育科技有限公司 | A kind of extraction method, device, the platform of mathematical problem knowledge point and solution approach |
CN110516697A (en) * | 2019-07-15 | 2019-11-29 | 清华大学 | Statement verification method and system based on evidence figure polymerization and reasoning |
CN110555143A (en) * | 2018-03-27 | 2019-12-10 | 北京世纪好未来教育科技有限公司 | Question automatic answering method and computer storage medium |
US10628743B1 (en) | 2019-01-24 | 2020-04-21 | Andrew R. Kalukin | Automated ontology system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006215847A (en) * | 2005-02-04 | 2006-08-17 | Hitachi Systems & Services Ltd | Inference method and inference device |
CN101414310A (en) * | 2008-10-17 | 2009-04-22 | 山西大学 | Method and apparatus for searching natural language |
CN104882040A (en) * | 2015-05-15 | 2015-09-02 | 陈爱秋 | Intelligent system applied in Chinese language teaching |
-
2017
- 2017-04-24 CN CN201710272107.5A patent/CN107169043A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006215847A (en) * | 2005-02-04 | 2006-08-17 | Hitachi Systems & Services Ltd | Inference method and inference device |
CN101414310A (en) * | 2008-10-17 | 2009-04-22 | 山西大学 | Method and apparatus for searching natural language |
CN104882040A (en) * | 2015-05-15 | 2015-09-02 | 陈爱秋 | Intelligent system applied in Chinese language teaching |
Non-Patent Citations (1)
Title |
---|
刘显玲: "初中数学错误模型研究及其在自动判卷系统中的应用", 《万方数据库》 * |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107992482A (en) * | 2017-12-26 | 2018-05-04 | 科大讯飞股份有限公司 | Mathematics subjective item answers the stipulations method and system of step |
CN108109453A (en) * | 2017-12-29 | 2018-06-01 | 广州大学 | Work correction device |
CN108255812A (en) * | 2018-01-16 | 2018-07-06 | 西南大学 | Calculus aptitude tests topic intelligent generation method based on semantic tagger |
CN108255812B (en) * | 2018-01-16 | 2021-03-23 | 西南大学 | Intelligent generation method of calculus capability test questions based on semantic annotation |
CN108182177A (en) * | 2018-01-24 | 2018-06-19 | 谢德刚 | A kind of mathematics knowledge-ID automation mask method and device |
CN110555143A (en) * | 2018-03-27 | 2019-12-10 | 北京世纪好未来教育科技有限公司 | Question automatic answering method and computer storage medium |
CN110555143B (en) * | 2018-03-27 | 2021-10-15 | 北京世纪好未来教育科技有限公司 | Question automatic answering method and computer storage medium |
CN109783775A (en) * | 2019-01-18 | 2019-05-21 | 广东小天才科技有限公司 | A kind of method and system for the content marking user's corpus |
US10628743B1 (en) | 2019-01-24 | 2020-04-21 | Andrew R. Kalukin | Automated ontology system |
CN109902160A (en) * | 2019-01-30 | 2019-06-18 | 华中师范大学 | The method and system that circuit topic is answered automatically |
CN109902160B (en) * | 2019-01-30 | 2023-07-25 | 华中师范大学 | Method and system for automatically solving circuit questions |
CN109918514A (en) * | 2019-03-21 | 2019-06-21 | 江西风向标教育科技有限公司 | A kind of extraction method, device, the platform of mathematical problem knowledge point and solution approach |
CN110516697A (en) * | 2019-07-15 | 2019-11-29 | 清华大学 | Statement verification method and system based on evidence figure polymerization and reasoning |
CN110516697B (en) * | 2019-07-15 | 2021-08-31 | 清华大学 | Evidence graph aggregation and reasoning based statement verification method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107169043A (en) | A kind of knowledge point extraction method and system based on model answer | |
Dror et al. | Replicability analysis for natural language processing: Testing significance with multiple datasets | |
CN105528437B (en) | A kind of question answering system construction method extracted based on structured text knowledge | |
CN107590133A (en) | The method and system that position vacant based on semanteme matches with job seeker resume | |
CN104035975B (en) | It is a kind of to realize the method that remote supervisory character relation is extracted using Chinese online resource | |
CN108664599B (en) | Intelligent question-answering method and device, intelligent question-answering server and storage medium | |
Tiddi et al. | Dedalo: Looking for clusters explanations in a labyrinth of linked data | |
CN112667794A (en) | Intelligent question-answer matching method and system based on twin network BERT model | |
CN106815252A (en) | A kind of searching method and equipment | |
Saleena et al. | Using concept similarity in cross ontology for adaptive e-Learning systems | |
CN106991161A (en) | A kind of method for automatically generating open-ended question answer | |
CN104133817A (en) | Online community interaction method and device and online community platform | |
CN107679110A (en) | The method and device of knowledge mapping is improved with reference to text classification and picture attribute extraction | |
CN110111010B (en) | Question and answer task allocation method and system based on crowd-sourcing network | |
CN106960001A (en) | A kind of entity link method and system of term | |
CN112597316A (en) | Interpretable reasoning question-answering method and device | |
CN113886567A (en) | Teaching method and system based on knowledge graph | |
CN111177399A (en) | Knowledge graph construction method and device | |
CN104462207B (en) | The multiple clips education resource mask method of Based on Distributed academic environment | |
Zubiaga et al. | Content-based clustering for tag cloud visualization | |
CN115204156A (en) | Keyword extraction method and device | |
CN111931516A (en) | Text emotion analysis method and system based on reinforcement learning | |
CN116561538A (en) | Question-answer scoring method, question-answer scoring device, electronic equipment and storage medium | |
CN103593334B (en) | A kind of method and system for being used to judge emotional degree of text | |
CN117216221A (en) | Intelligent question-answering system based on knowledge graph and construction method |
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 | ||
TA01 | Transfer of patent application right | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20190805 Address after: Room 11H, Building No. 33 Zizhuyuan Road, Haidian District, Beijing 100089 Applicant after: Lin Hui Address before: 610000 Fifth Floor, Building No. 5, Tianfu New Valley, 399 West Section of Fucheng Avenue, Chengdu High-tech Zone, Sichuan Province Applicant before: Chengdu foresight nehology Science and Technology Ltd. |
|
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170915 |