CN109190129A - A kind of multilingual translation quality evaluation engine based near synonym knowledge mapping - Google Patents
A kind of multilingual translation quality evaluation engine based near synonym knowledge mapping Download PDFInfo
- Publication number
- CN109190129A CN109190129A CN201810997778.2A CN201810997778A CN109190129A CN 109190129 A CN109190129 A CN 109190129A CN 201810997778 A CN201810997778 A CN 201810997778A CN 109190129 A CN109190129 A CN 109190129A
- Authority
- CN
- China
- Prior art keywords
- knowledge
- translation
- word
- evaluation
- knowledge mapping
- 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
- 238000013519 translation Methods 0.000 title claims abstract description 57
- 238000013507 mapping Methods 0.000 title claims abstract description 38
- 238000013441 quality evaluation Methods 0.000 title description 13
- 238000011156 evaluation Methods 0.000 claims abstract description 38
- 238000004458 analytical method Methods 0.000 claims abstract description 33
- 238000012360 testing method Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000000034 method Methods 0.000 claims description 29
- 238000001914 filtration Methods 0.000 claims description 6
- 238000002360 preparation method Methods 0.000 claims description 2
- 230000014616 translation Effects 0.000 description 46
- 230000006870 function Effects 0.000 description 5
- 238000007477 logistic regression Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000002688 persistence Effects 0.000 description 1
- 238000001303 quality assessment method Methods 0.000 description 1
- 238000000528 statistical test Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/51—Translation evaluation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Machine Translation (AREA)
Abstract
The present invention proposes a kind of multilingual translation QA system based near synonym knowledge mapping, and the evaluation system includes source languages input module, analytic unit, evaluation component and knowledge library component；The knowledge library component includes the near synonym knowledge mapping knowledge base that sustainable training updates；Analysis result is inputted the evaluation component, to obtain the evaluation score of tester based on the test result that the knowledge base block analysis tester provides for the corpus to be translated of source languages input module input by the analytic unit.The invention also includes corresponding evaluation methods.
Description
Technical field
The present invention relates to translation quality evaluations, more particularly to a kind of turned over based on the multilingual of near synonym knowledge mapping
Translate quality evaluating method and system.
Background technique
Currently, the written translation ability of a people is evaluated, marking evaluation mainly is carried out to it by various test modes.Example
Such as, the examination methods of marking combined using subjective item and objective item.For objective item, candidate only needs to select just
True candidate item can score, this partial evaluation work can be completed by machine statistical test；For subjective item, by
It is different in the translation that different translators provides same entry to be translated, and the Key for Reference that the person that usually do not set a question provides has
Limit easily generates erroneous judgement if counted using simple machine, as it is likely that there is each not phase that different translators provide
With translation result be possible to be all different with Key for Reference, but the case where be correct option.
At this point, usually introducing the means manually marked examination papers.But when the enormous amount of paper, if every portion paper is adopted
Manually mark examination papers, then will be obviously improved test job amount, increases testing cost.
And in actual scene, when interpreter translates and examines interpreter, can all there be the translation matter to interpreter
The process evaluated is measured, this process is also all to have corresponding GUIDED TRANSLATION teacher to carry out, unusual labor intensive, while again
Very strong subjectivity.
Summary of the invention
To solve the problems, such as that existing translation ability assessment and translation quality assessment exist, the invention proposes based on close
The translation quality evaluation of adopted word knowledge mapping, using the knowledge mapping of a set of multilingual near synonym as basic knowledge base,
A set of assessment machine, " Evaluation Machine " are used on the basis of it, abbreviation EM carries out the translation quality of assessment interpreter.
In the first aspect of the invention, a kind of multilingual translation quality evaluation system based near synonym knowledge mapping is proposed
System, the evaluation system includes source languages input module, analytic unit, evaluation component and knowledge library component；The knowledge base
Component includes the near synonym knowledge mapping knowledge base that sustainable training updates；The analytic unit is based on the knowledge library component
The test result that analysis tester provides for the corpus to be translated of source languages input module input, will be described in analysis result input
Evaluation component, to obtain the evaluation score of tester；
The knowledge mapping knowledge base includes the similarity distance score of near synonym or related term；The evaluation component is based on described
Knowledge mapping knowledge base and model answer are fitted marking, obtain the translation quality score of tester.
When specific implementation, can persistence maintenance a set of automatically updated multilingual near synonym knowledge mapping in advance, this knowledge
Map includes the similarity distance score of near synonym or related term；The knowledge library component is based on constructing on the basis of word2vec
The distance metrics of near synonym and relative words.
When interpreter has been connected to one section of translation corpus, referred to as source languages corpus, object language, while model answer have been translated into
It can prepare several parts of model answers of source corpus, it is independent good that these answers are prepared in advance by teacher, ensure that the correctness of translation；
The i.e. described model answer is the translation result for corresponding to the corpus to be translated of preparation in advance；
Preferably, the model answer is more parts, every part of source independence.
Then, the analytic unit is segmented and is filtered to the test result, obtains key evaluating word.
Specifically, analytic unit can segment the translation corpus of interpreter, it then will do it a filtering, filter
Word out can be critical evaluation word, for example the ratio translated for the semantic word for having central role and certain professional domain needs is calibrated
Word；
Preferably, above system further includes that knowledge library component updates engine, the output knot based on analytic unit and/or evaluation component
Fruit is updated the knowledge mapping knowledge base.Based on this point, what is obtained is that one kind can be known with the near synonym of continuous updating
Know map knowledge base, in this, as the knowledge base basis of evaluation, ensure that vocabulary can closely follow the development of upper language, be unlikely to fall
Afterwards；Meanwhile this knowledge base constructs the distance metric of near synonym and relative words on the basis of word2vec, ensure that vocabulary
Between relationship degree measurement, overcome original near synonym only whether the obstacle of relationship, while improving the fining of this method
Degree.
It is worth noting that the output based on analytic unit and/or evaluation component is as a result, to the knowledge mapping knowledge base
It is updated, not only includes actively updating, further include passively updating.It actively updates, refers to the evaluation score according to output to determine
It is fixed whether to update corpus, knowledge base；It is passive to update, then it can choose and regularly updated from internet corpus.
Different from the prior art, the present invention can also be set when the output result of analytic unit and/or evaluation component meets
When predetermined condition, the update of manual feedback prompt is realized.This point illustrates emphasis in specific embodiments of the present invention part.
Using this mechanism, erroneous judgement problem that may be present can be assessed to avoid simple equipment analysis and machine.
In the second aspect of the invention, a kind of multilingual translation quality evaluation based near synonym knowledge mapping is proposed
Method, the method includes source languages input step, target language analytical procedure, target language appraisal procedures；
The source languages input step, inputs corpus to be translated, and personnel to be measured are based on the corpus to be translated and provide test result composition
Target language；
The target language analytical procedure, analyzes the test result, provides critical evaluation word；
The target language appraisal procedure is based on the key evaluating word, comments the translation quality of the personnel to be measured
Estimate；
The target language analytical procedure, based on the near synonym knowledge mapping knowledge base that sustainable training updates, to the test
As a result it carries out participle and filtration treatment, the distance metric based near synonym and relative words obtains the key evaluating word.
The target language appraisal procedure is fitted marking based on the knowledge mapping knowledge base and model answer, obtains
The translation quality score of personnel to be measured out.
The model answer is from a variety of different translation engines；And/or different human translations.
Preferably, further include that feedback updates step, impose a condition when the translation quality score of the personnel to be measured meets
When, the knowledge mapping knowledge base is updated.
It, can be to avoid it should be pointed out that knowledge library component of the present invention updates engine, feedback update step
Simple equipment analysis and machine assess erroneous judgement problem that may be present.In specific embodiments of the present invention part, by emphasis
This is illustrated.
Detailed description of the invention
Fig. 1 is the architecture diagram of translation quality evaluation of the invention
Fig. 2 is translation quality evaluation method flow chart of the invention
Fig. 3 is evaluation component (EM) core architecture figure in translation quality evaluation of the invention
Fig. 4 is a kind of membership credentials figure of near synonym knowledge base
Fig. 5 is the instance graph for calculating word and moving distance
Specific embodiment
It referring to Fig.1, is a specific architecture diagram of multilingual translation QA system of the invention, the evaluation system
Including the input of source languages, analysis module, evaluation component (EM) and knowledge library component, (corpus constantly accumulated is cyclically updated closely
Adopted word knowledge base)；The knowledge library component includes the near synonym knowledge mapping knowledge base that sustainable training updates；The analysis mould
Block, the test knot provided based on the knowledge base block analysis tester for the corpus to be translated of source languages input module input
Analysis result is inputted the evaluation component, to obtain the evaluation score of tester by fruit.
In the concrete realization, evaluation component is evaluation automatic machine (EM), can be passed through on the analysis foundation of analysis module
Model obtains the score of translation quality, this EM is based on machine learning and is fitted to obtain, and is the mould obtained by the training of a large amount of corpus
Type, model ensure that the fairness of evaluation, abandon the subjectivity of special messenger's evaluation；
It is translation quality evaluation method flow chart of the invention of the invention, including source languages input step, target referring to Fig. 2
Languages analytical procedure, target language appraisal procedure；
The source languages input step, inputs corpus to be translated, and personnel to be measured are based on the corpus to be translated and provide test result composition
Target language；
The target language analytical procedure, analyzes the test result, provides critical evaluation word；
The target language appraisal procedure is based on the key evaluating word, comments the translation quality of the personnel to be measured
Estimate；
The target language analytical procedure, based on the near synonym knowledge mapping knowledge base that sustainable training updates, to the test
As a result it carries out participle and filtration treatment, the distance metric based near synonym and relative words obtains the key evaluating word.
The target language appraisal procedure is fitted marking based on the knowledge mapping knowledge base and model answer, obtains
The translation quality score of personnel to be measured out.
When specific implementation, the above process be may be summarized to be:
1) safeguard that a set of automatically updated multilingual near synonym knowledge mapping, this knowledge mapping include near synonym or correlation
The similarity distance score of word；
2) when interpreter has been connected to one section of translation corpus, referred to as source corpus, object language has been translated into, while model answer can be quasi-
Several parts of model answers of standby source corpus, it is independent good that these answers are prepared in advance by teacher, ensure that the correctness of translation；
3) analysis module can segment the translation corpus of interpreter, then will do it a filtering, filtering the word obtained can be
Critical evaluation word, such as word that semanteme has the word of central role and certain professional domain to need the ratio translated calibrated；
4) result of analysis module is sent into EM, EM can be carried out near synonym knowledge mapping and the common base of model answer
Fitting marking, obtains the translation quality score of interpreter.
As another innovative point of the invention, method of the present invention includes that feedback updates step, when it is described to
When the translation quality score of survey personnel meets setting condition, the knowledge mapping knowledge base is updated.Update herein is not
The problems such as only conventional corpus updates, and can also save outstanding translation result, and equipment analysis bring is avoided to judge by accident.
Inventor has found that computer based analytic unit can not be forever quasi- in the long-term quality evaluation course of work
Really, automatic assessment machine (EM) essence based on machine learning is mechanical, and the assessment result provided has no idea to embody literary grace.
And language translation is then the infinite process of a potentiality, in literature, it is likely that occur, tester gives one and exquisite turns over
It translates as a result, still it does not include key evaluating word that any model answer occurs, at this point, appraisal result may directly score is
Zero, so as to cause erroneous judgement.
One simply example is as follows:
For example, the corpus to be translated of input is
(T1)I look for what I miss, I know not what it is, I feel so sad, so
drear, so lonely, without cheer.
Its model answer may be:
(A1) I finds my miss, I does not know that it is, I feels such sadness, so cloudy, so lonely, does not have
It hails.
(A2) I finds what I missed, and it is what that I, which does not know, I feels to be hard hit, so dull, so lonely, does not have
It hails.
But as a high-caliber translator (personnel to be measured), which show following test results:
(A3) it seeks and looks for, it is desolate, it is heartbreaking sad.
But above-mentioned more graceful answer, regardless of analytic unit segments it, how to be filtered, how
Near synonym analysis etc. is carried out, the impossible primary election in model answer of the crucial part of speech evaluating word obtained, at this point, if following blindly machine
Device engine then necessarily leads to erroneous judgement, and therefore, feedback updates step and has an effect, and manual feedback module detects that the result is
More preferably as a result, being saved into EM module and knowledge mapping knowledge base.Only consider that score is higher different from the prior art
Situation, in the present invention, cause settings condition that feedback updates not only can be tester score it is higher (higher than certain threshold value),
It can also be that the score of tester is obvious relatively low (even 0).
It is the evaluation group in translation quality evaluation of the invention in a specific translation quality evaluation referring to Fig. 3
Part (EM) core architecture figure.Wherein, the value of logistic regression function F output is exactly the translation quality score of interpreter.
Firstly, it is necessary to which, it is noted that comparative diagram 3 and Fig. 1 are it is found that logistic regression function can be equivalent in Fig. 1 in Fig. 3
The combination of analysis module and EM module；
But the present invention may be designed in, interpreter's answer in Fig. 3 has been that treated by the analysis module in Fig. 1
Then efficient matchings input results obtain scoring into logistic regression function F.
It is specifically described its implementation process and related concept below.
Near synonym knowledge base, this knowledge base press following relational organization, wherein ellipse is word node, the line between word node
Relatedness metric is 0.762 between its relatedness metric, such as aircraft and helicopter；It specifically can refer to Fig. 4.
Based on the sentence similarity evaluation module of " word shifting distance ", module is accomplished by
The 1 near synonym knowledge base by the training of a large amount of corpus is mainly used to adjust the distance metric between word, guarantees the accurate of distance
Property；
2 will train machine learning method one fitting function of training an of logistic regression simultaneously, be denoted as F；
The similarity distance (word-based shifting distance) that the answer and model answer of 3 calculating interpreters is shown in, then presses each model answer
It is inputted according to fitting function F, F can export a value, this value is exactly interpreter's translation quality score；Calculation formula is,
InFor i, distance corresponding to two words of j near synonym knowledge mapping can thus calculate score.
Word moves distance
It is to do basis with word2vec that word, which moves distance, the method to calculate the similitude between 2 documents.The basic think of of word shifting distance
Want that calculating term vector corresponding to any two word in two documents asks Euclidean distance then weighted sum again, is probably this
The form of sample:, whereinFor i, (nearly justice is used herein in the Euclidean distance of term vector corresponding to two words of j
Distance in word knowledge mapping).Here key is exactly the calculating of Ti, and specific calculate sees Fig. 5.
There are two document (document1- document 2) in Fig. 5, and after removing stop words, every document is only left 4
Word, we seek to compare the similarity between two documents with this four words.Herein, it will be assumed that ' Obama ' this
Weight of the word in document 1 is that 0.5(can be calculated simply with word frequency or TFIDF), then due to ' Obama '
' president ' the very high weight very high with the similarity of ' president ', then we can be moved to by ' Obama ',
It is assumed that be 0.4, other words are due to distant with the distance of ' Obama ' in document 2, so smaller weight can be assigned to.
Here constraint is, by some word ii in document 1 be moved to the sum of weight of each word in document 2 should in document 1
This word ii weight it is equal, i.e., the weight (0.5) of oneself is given each word in document 2 by ' Obama '.Equally, literary
Some word jj in shelves 2, which is received, should be equal to word jj in document 2 by the sum of the weight that each word in document 1 is flowed into
Weight.
This cost is acquired after lower bound minimizes, word in all document a can be acquired and be transferred to text
The most short total distance of word, represents the similarity between two documents in shelves b.Word moves the computation complexity of distance algorithm by excellent
It can achieve O (p2) after changing, wherein p is the number of not repetitor in document.
In general, innovative point of the invention includes at least:
1) the near synonym knowledge mapping knowledge base that a set of sustainable training updates；
2) analysis module is translated, the output of this analysis can supply EM；
3) evaluation automatic machine (EM) can obtain the score of translation quality by model on the analysis foundation of analysis module.
4) conventional better quality translation result and unconventional exquisite translation result have been comprehensively considered, have been unlikely to allow simple
Mechanical machine translation generate erroneous judgement.
Claims (10)
1. a kind of multilingual translation QA system based near synonym knowledge mapping, the evaluation system includes that source languages are defeated
Enter component, analytic unit, evaluation component and knowledge library component；The knowledge library component includes the nearly justice that sustainable training updates
Word knowledge mapping knowledge base；The analytic unit is directed to source languages input group based on the knowledge base block analysis tester
Analysis result is inputted the evaluation component, to obtain tester's by the test result that the corpus to be translated of part input provides
Evaluate score；
It is characterized by:
The knowledge mapping knowledge base includes the similarity distance score of near synonym or related term；The evaluation component is based on described
Knowledge mapping knowledge base and model answer are fitted marking, obtain the translation quality score of tester.
2. the system as claimed in claim 1, wherein the analytic unit is segmented and filtered to the test result, is obtained
Key evaluating word out.
3. system as claimed in claim 1 or 2, wherein the knowledge library component is based on constructing closely on the basis of word2vec
The distance metric of adopted word and relative words.
4. the system as claimed in claim 1, the model answer is the translation for corresponding to the corpus to be translated of preparation in advance
As a result.
5. the system as described in claim 4 or 1, the model answer is more parts, every part of source independence.
6. the system as claimed in claim 1 further comprises that knowledge library component updates engine, based on analytic unit and/or comments
The output of component is estimated as a result, being updated to the knowledge mapping knowledge base.
7. a kind of multilingual translation quality evaluating method based near synonym knowledge mapping, the method includes the inputs of source languages to walk
Suddenly, target language analytical procedure, target language appraisal procedure；
The source languages input step, inputs corpus to be translated, and personnel to be measured are based on the corpus to be translated and provide test result composition
Target language；
The target language analytical procedure, analyzes the test result, provides key evaluating word；
The target language appraisal procedure is based on the key evaluating word, comments the translation quality of the personnel to be measured
Estimate；
It is characterized by: the target language analytical procedure, based on the near synonym knowledge mapping knowledge base that sustainable training updates,
Participle is carried out to the test result and filtration treatment, the distance metric based near synonym and relative words obtain the pass
Keyness evaluating word.
8. the method for claim 7, the target language appraisal procedure is based on the knowledge mapping knowledge base and standard
Answer is fitted marking, obtains the translation quality score of personnel to be measured.
9. method according to claim 8, wherein the model answer is from a variety of different translation engines；And/or it is different
Human translation.
10. further including such as the described in any item methods of claim 7-9, feedback updates step, when the translation of the personnel to be measured
When quality score meets setting condition, the knowledge mapping knowledge base is updated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810997778.2A CN109190129A (en) | 2018-08-31 | 2018-08-31 | A kind of multilingual translation quality evaluation engine based near synonym knowledge mapping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810997778.2A CN109190129A (en) | 2018-08-31 | 2018-08-31 | A kind of multilingual translation quality evaluation engine based near synonym knowledge mapping |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109190129A true CN109190129A (en) | 2019-01-11 |
Family
ID=64916603
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810997778.2A Pending CN109190129A (en) | 2018-08-31 | 2018-08-31 | A kind of multilingual translation quality evaluation engine based near synonym knowledge mapping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109190129A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858029A (en) * | 2019-01-31 | 2019-06-07 | 沈阳雅译网络技术有限公司 | A kind of data preprocessing method improving corpus total quality |
| CN111753560A (en) * | 2020-06-28 | 2020-10-09 | 语联网（武汉）信息技术有限公司 | Self-adaptive translation system under parallel multi-source input mode |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101923540A (en) * | 2010-07-20 | 2010-12-22 | 陈洁 | Language translation quality auditing method |
| CN107122467A (en) * | 2017-04-26 | 2017-09-01 | 努比亚技术有限公司 | The retrieval result evaluation method and device of a kind of search engine, computer-readable medium |
| CN107274738A (en) * | 2017-06-23 | 2017-10-20 | 广东外语外贸大学 | Chinese-English translation teaching points-scoring system based on mobile Internet |
-
2018
- 2018-08-31 CN CN201810997778.2A patent/CN109190129A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101923540A (en) * | 2010-07-20 | 2010-12-22 | 陈洁 | Language translation quality auditing method |
| CN107122467A (en) * | 2017-04-26 | 2017-09-01 | 努比亚技术有限公司 | The retrieval result evaluation method and device of a kind of search engine, computer-readable medium |
| CN107274738A (en) * | 2017-06-23 | 2017-10-20 | 广东外语外贸大学 | Chinese-English translation teaching points-scoring system based on mobile Internet |
Non-Patent Citations (1)
| Title |
|---|
| MATT J.KUSNER等: "From Word Embeddings To Document Distances", 《JOURNAL OF MACHINE LEARNING RESEARCH》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858029A (en) * | 2019-01-31 | 2019-06-07 | 沈阳雅译网络技术有限公司 | A kind of data preprocessing method improving corpus total quality |
| CN109858029B (en) * | 2019-01-31 | 2023-02-10 | 沈阳雅译网络技术有限公司 | Data preprocessing method for improving overall quality of corpus |
| CN111753560A (en) * | 2020-06-28 | 2020-10-09 | 语联网（武汉）信息技术有限公司 | Self-adaptive translation system under parallel multi-source input mode |
| CN111753560B (en) * | 2020-06-28 | 2024-05-10 | 语联网（武汉）信息技术有限公司 | Self-adaptive translation system in parallel multi-source input mode |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pavlick et al. | Simple PPDB: A paraphrase database for simplification | |
| Smith et al. | " I'm sorry to hear that": Finding New Biases in Language Models with a Holistic Descriptor Dataset | |
| De Belder et al. | A dataset for the evaluation of lexical simplification | |
| Teruya et al. | Halliday in relation to language comparison and typology | |
| EP3100174A1 (en) | Method for automatically detecting meaning and measuring the clearness of text | |
| CN105912629A (en) | Intelligent question and answer method and device | |
| Shutova | Metaphor identification as interpretation | |
| KR20150082783A (en) | Semantic Frame Operating Method Based on Text Big-data and Electronic Device supporting the same | |
| CN109190129A (en) | A kind of multilingual translation quality evaluation engine based near synonym knowledge mapping | |
| Berzak et al. | Contrastive analysis with predictive power: Typology driven estimation of grammatical error distributions in ESL | |
| Zhou et al. | Development of WRRL: A new computerized measure of the reflecting/reorganizing function | |
| Bagaria et al. | An intelligent system for evaluation of descriptive answers | |
| CN106803216A (en) | The aspiration commending system and method in a kind of adaptation Zhejiang Province | |
| CN106502988B (en) | A kind of method and apparatus that objective attribute target attribute extracts | |
| Bracewell et al. | A tiered approach to the recognition of metaphor | |
| El et al. | Intelligent Arabic text to arabic sign language translation for easy deaf communication | |
| CN109118109A (en) | Quality evaluation based on ETM | |
| Siklósi et al. | Resolving abbreviations in clinical texts without pre-existing structured resources | |
| Huang et al. | A robust estimation scheme of reading difficulty for second language learners | |
| Bartolini et al. | From Synsets to Videos: Enriching ItalWordNet Multimodally. | |
| Shen et al. | Sliqa-i: Towards cold-start development of end-to-end spoken language interface for question answering | |
| Corrigan et al. | An ontolological treatment of clinical prediction rules implementing the alvarado score | |
| Hou et al. | Language change in Chinese political discourse based on the relationship between sentence and clause | |
| Lichouri et al. | An Arabic multi-domain spoken language understanding system | |
| Le et al. | Joint learning of frequency and word embeddings for multilingual readability assessment |
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 |
Application publication date: 20190111 |
|
| RJ01 | Rejection of invention patent application after publication |