CN106021239A - Method for real-time evaluation of translation quality - Google Patents
Method for real-time evaluation of translation quality Download PDFInfo
- Publication number
- CN106021239A CN106021239A CN201610272936.9A CN201610272936A CN106021239A CN 106021239 A CN106021239 A CN 106021239A CN 201610272936 A CN201610272936 A CN 201610272936A CN 106021239 A CN106021239 A CN 106021239A
- Authority
- CN
- China
- Prior art keywords
- translation
- engine
- statement
- user
- quality
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/40—Processing or translation of natural language
- G06F40/51—Translation evaluation
-
- 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
- G06F40/58—Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
Abstract
The invention discloses a method for real-time evaluation of translation quality. The method comprises the steps that a statement initiated by a user is received according to the user's translation request; multiple translation engines receive the statement initiated by the user; according to statistic results of total quantity of translation times of each engine as well as frequencies for the user to collect and play the statement, a weighted utilization frequency of translation results of each translation engine, and a per capita utilization frequency are calculated and taken as a quality score; the proportion occupied by the quality score in the quality score sum of all the engines is taken as a probability; the translation engine with the highest probability is taken as the optimal translation engine for the statement request; and the obtained optimal translation engine is taken as a preferential engine for the statement request, the translation engine is called to translate the statement requested by the user, and finally a translation result is obtained. The method disclosed by the invention has the advantages that the automatic and real-time quality evaluation is carried out to the translation quality of the multiple engines; and optimal translation engine results can be provided for different statements in need of translation.
Description
Technical field
The present invention relates to translation quality and evaluate field, specifically a kind of translation quality real-time estimating method.
Background technology
People are when using translation tool, it is common that realized by a kind of translation engine, and this translation result of artificial cognition
Accuracy, if feeling, translation result is not ideal enough, then can attempt changing a kind of translation engine and attempt, until cannot obtain again
Till more excellent result.The mode of this manual switching translation engine, it is thus achieved that the efficiency of optimum translation result is low, the longest, especially
It cannot meet the demand of real time translation.
So, integrated multiple translation engines, provide the user the higher translation result of coverage, it is possible to part solves above-mentioned skill
Art problem, is particularly suited for overseas trip user.In order to provide the translation result of high-quality to overseas trip user, promote translation and cover
Cover degree, the integrated multiple translation engines of meeting, it is ensured that the coverage of translation, but same statement to be translated, limited is commonly used
The use of statement, can ask expert to evaluate, and selects and determines optimum translation result, and this mode obviously cannot be answered
To complicated translation demand.The mode of existing experts' evaluation, maximum shortcoming is to accomplish quick real-time response, right
Translating statement in magnanimity, manpower evaluation cost is high, it is impossible to accomplish the degree of automatization and quantization, also cannot sustained improvement evaluation
Mode.Another way is that each translation engine is sampled quality evaluation, evaluates the synthesized translation quality of certain engine.
But cannot accomplish to be accurate to the translation quality of different statement.
More intractable, for real-time translation on line, how the translation result of the multiple engine of quantitative evaluation, is to obtain
Excellent translation result have to solve the technical problem that, this has just become a technical problem the most urgent.Prior art is often led to
Crossing analysis linguistic structure to solve number of patent application disclosed in this technical problem, such as State Intellectual Property Office is
The application for a patent for invention machine translation method of 201410461334.9 and system, but this cannot adapt to different linguistic context, field equally
Close and the language convention etc. of user, therefore cannot solve above-mentioned technical problem.
Summary of the invention
In view of this, the technology that cannot obtain optimum translation result efficiently that the present invention is directed to the existence of above-mentioned prior art is asked
Topic, it is provided that a kind of integrated multiple translation engines the translation quality real-time estimating method of real-time online assessment, to obtain optimum
Translation result.
The technical solution of the present invention is to provide the translation quality real-time estimating method of a kind of following steps, including following step
Rapid:
Step 1: ask according to the translation of user, receive the statement that user is initiated;
Step 2: integrated multiple translation engines all receive described Client-initiated statement, according to adding up each of different engine
From total translation number of times and user collected this statement, play the frequency of this statement, calculate the translation result of corresponding translation engine
Weighting frequency of usage and calculate frequency of usage per capita and divide as quality, and divide the quality accounting for all engines to divide to add sum with quality
Ratio is probability, and the highest translation engine of probability is then as the optimum translation engine of this statement requests;
Step 3: the above-mentioned optimum translation engine obtained preferentially is used engine as ask this statement, dispatches this translation
The statement that user is asked translated by engine, obtains translation result, returns to user.
Using above step, the present invention compared with prior art, has the advantage that the employing present invention, initiates for user
Particular statement, each translation engine is not quite similar, by calculating the weighting frequency of usage of translation result of corresponding translation engine
Divide as quality with calculating frequency of usage per capita, as to different translation engines for the evaluation of the translation result of this statement, with
The person of having higher rating, as the preferential translation engine used, completes the selection to this statement optimum translation engine, thus obtains optimum
Translation result, returns to user.Invention introduces the frequency as Appreciation gist, it is not necessary to the linguistic structure of anolytic sentence, energy
Enough fast and effeciently obtain optimum translation result, by this systematic method, the translation quality to multiple engines, carry out
The real-time quality evaluation of automatization, for different translation statements, provides the translation engine result of optimum.
As preferably, after completing step 3, update the translation of this statement under described optimum translation engine, user initiated
The frequency, the user mentioning the request of this statement translation is using the follow-up behavior of translation result, is being sent to quality evaluation as daily record
Data base, for follow-up evaluation with to the dynamic corrections selecting optimum translation engine.
As preferably, quality assessment data storehouse is according to the log information reported, and gathers the positional information of user to judge user
Location and the most abroad, in conjunction with user's geographical location circumstances as weighing the cofactor of translation quality, just can setting district
Territory scope, obtains the optimum translation engine of this regional extent.After superposition positional information, more can accurately reflect translation engine
Translation quality, because the quality of translation is often relevant with the use scene of statement.
As preferably, for the identical translation request of different user, different user uses different engine to obtain translation result, and
By analyzing and processing subsequent user to translation result in different time, the occupation mode of different geographical and the frequency, quantitative evaluation is not
With the translation engine translation quality to this statement, thus provide the best translation result of this statement for other users.By this
Plant comprehensive analysis and the mode of assessment, it is possible to obtain more excellent translation result.
Accompanying drawing explanation
Fig. 1 is the step schematic diagram of translation quality real-time estimating method;
Detailed description of the invention
The invention will be further described with specific embodiment below in conjunction with the accompanying drawings, but the present invention is not restricted to these enforcement
Example.
The present invention contains any replacement, amendment, equivalent method and scheme made in the spirit and scope of the present invention.In order to
Make the public that the present invention to be had to understand thoroughly, in present invention below preferred embodiment, concrete details is described in detail, and right
The description not having these details for those skilled in the art can also understand the present invention completely.
As it is shown in figure 1, a kind of translation quality real-time estimating method of the present invention, comprise the following steps:
Step 1: ask according to the translation of user, receive the statement that user is initiated;
Step 2: integrated multiple translation engines all receive described Client-initiated statement, according to adding up each of different engine
From total translation number of times and user collected this statement, play the frequency of this statement, calculate the translation result of corresponding translation engine
Weighting frequency of usage and calculate frequency of usage per capita and divide as quality, and divide the quality accounting for all engines to divide to add sum with quality
Ratio is probability, and the highest translation engine of probability is then as the optimum translation engine of this statement requests;
In step 2, can be with setting data amount, the threshold value of quality, after reaching, do not return optimum translation engine further according to probability,
Because having obtained reliable result, then need not repeat every time, improving response speed.
Step 3: the above-mentioned optimum translation engine obtained preferentially is used engine as ask this statement, dispatches this translation
The statement that user is asked translated by engine, obtains translation result, returns to user.
After completing step 3, update the translation frequency of this statement under described optimum translation engine, user initiated, mention
The user of this statement translation request is using the follow-up behavior of translation result, is sent to quality assessment data storehouse as daily record, uses
In follow-up evaluation with to the dynamic corrections selecting optimum translation engine.
Quality assessment data storehouse is according to the log information reported, and gathers the positional information of user to judge user location and to be
No abroad, in conjunction with user's geographical location circumstances as weigh translation quality cofactor, just can setting regions scope,
Optimum translation engine to this regional extent.After superposition positional information, more can accurately reflect the translation quality of translation engine,
Because the quality of translation is often relevant with the use scene of statement, scene and geographical position is used often to determine use habit,
Also different communicative habits is existed for.
For the identical translation request of different user, different user uses different engine to obtain translation result, and by analysis at
Manage subsequent user to translation result in different time, the occupation mode of different geographical and the frequency, quantitative evaluation difference translation engine
Translation quality to this statement, thus the best translation result of this statement is provided for other users.By this comprehensive analysis
Mode with assessment, it is possible to obtain more excellent translation result.
It should be noted that quality assessment data storehouse mentioned above refers to the quality evaluation storehouse in Fig. 1.
The concrete principle steps of the present invention is as follows: in particular for being in the people of overseas travelling, initiate statement translation to client
Request, obtain its geographical location information, according to the use habit in this area, and combine the translation to respective request statement
The frequency and the follow-up behavior of user, provide the user the translation engine of optimum, it is thus achieved that optimum translation result, and many because of
Element comprehensive under, the selection of translation engine can be carried out dynamic corrections.
The more specifically application scenarios of the present invention is following (being realized by mobile terminal APP):
Step 1: client initiates the translation request of statement.
Step 2: translation engine this statement of dispatch request preferentially use engine.
Step 3: the receipts that quality evaluation storehouse is abroad carried out according to respective total translation number of times and the subsequent user of the different engine of statistics
Tibetan language sentence, play the frequency of statement, calculate the weighting frequency of usage of this translation result and calculate frequency of usage per capita and divide as quality,
And divide the quality accounting for all engines to divide the ratio adding sum as probability with quality, return to translation engine scheduling point corresponding the drawing of person
Hold up, and update lower this engine translation frequency to this statement.
Step 4: according to returning engine, the translation result of the corresponding engine of request.
Step 5: corresponding engine returns translation result and dispatches to translation engine.
Step 6: translation result is returned to client by translation engine scheduling.
Step 7: user uses the follow-up behavior of translation result in client, is sent to quality evaluation engine as daily record.
Step 8: the log information that quality evaluation engine reports according to client, calculates user location and the most abroad,
And add this daily record to quality evaluation storehouse, and calculate user's frequency of usage data
Below only preferred embodiment of the present invention is described, but is not to be construed as limitations on claims.The present invention
Being not only limited to above example, its concrete structure allows to change.In a word, all protections in independent claims of the present invention
In the range of the various changes made the most within the scope of the present invention.
Claims (4)
1. a translation quality real-time estimating method, comprises the following steps:
Step 1: ask according to the translation of user, receive the statement that user is initiated;
Step 2: integrated multiple translation engines all receive described Client-initiated statement, according to adding up each of different engine
From total translation number of times and user collected this statement, play the frequency of this statement, calculate the translation result of corresponding translation engine
Weighting frequency of usage and calculate frequency of usage per capita and divide as quality, and divide the quality accounting for all engines to divide to add sum with quality
Ratio is probability, and the highest translation engine of probability is then as the optimum translation engine of this statement requests;
Step 3: the above-mentioned optimum translation engine obtained preferentially is used engine as ask this statement, dispatches this translation
The statement that user is asked translated by engine, obtains translation result, returns to user.
After completing step 3, update the translation frequency of this statement under described optimum translation engine, user initiated, mention
The user of this statement translation request is using the follow-up behavior of translation result, is sent to quality assessment data storehouse as daily record, uses
In follow-up evaluation with to the dynamic corrections selecting optimum translation engine.
A kind of translation quality real-time estimating method the most according to claim 1, it is characterised in that: complete step 3
After, update the translation frequency of this statement under described optimum translation engine, user initiated, mention the request of this statement translation
User is using the follow-up behavior of translation result, is sent to quality assessment data storehouse as daily record, for follow-up evaluation and right
Select the dynamic corrections of optimum translation engine.
A kind of translation quality real-time estimating method the most according to claim 1 and 2, it is characterised in that: quality evaluation
Database root is according to the log information that reports, and gathers the positional information of user to judge user location and the most abroad, knot
Share family geographical location circumstances as weigh translation quality cofactor, just can setting regions scope, obtain this regional extent
Optimum translation engine.
A kind of translation quality real-time estimating method the most according to claim 1, it is characterised in that: for different user
Identical translation request, different user uses different engine to obtain translation result, and by analyzing and processing subsequent user to translation
Result is in different time, the occupation mode of different geographical and the frequency, the quantitative evaluation difference translation engine translation matter to this statement
Amount, thus the best translation result of this statement is provided for other users.By the way of this comprehensive analysis and assessment, energy
The translation result that enough acquisitions are more excellent.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610272936.9A CN106021239B (en) | 2016-04-29 | 2016-04-29 | A kind of translation quality real-time estimating method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610272936.9A CN106021239B (en) | 2016-04-29 | 2016-04-29 | A kind of translation quality real-time estimating method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106021239A true CN106021239A (en) | 2016-10-12 |
CN106021239B CN106021239B (en) | 2018-10-26 |
Family
ID=57081012
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610272936.9A Active CN106021239B (en) | 2016-04-29 | 2016-04-29 | A kind of translation quality real-time estimating method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106021239B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107894983A (en) * | 2017-11-08 | 2018-04-10 | 深圳市沃特沃德股份有限公司 | The switching method and device of engine |
CN107979856A (en) * | 2017-11-22 | 2018-05-01 | 深圳市沃特沃德股份有限公司 | Connect the method and apparatus of engine |
CN108319591A (en) * | 2018-02-05 | 2018-07-24 | 深圳市沃特沃德股份有限公司 | Realize the method, apparatus and speech translation apparatus of voiced translation |
CN109118109A (en) * | 2018-08-31 | 2019-01-01 | 传神语联网网络科技股份有限公司 | Quality evaluation based on ETM |
CN109214011A (en) * | 2018-09-19 | 2019-01-15 | 深圳市合言信息科技有限公司 | It is a kind of by user feedback come the cognitive engine selection strategy of self-perfection |
CN110502762A (en) * | 2019-08-27 | 2019-11-26 | 北京金山数字娱乐科技有限公司 | A kind of transcription platform and its management method |
CN110532574A (en) * | 2019-08-20 | 2019-12-03 | 语联网(武汉)信息技术有限公司 | MT engine selection method and device |
WO2020057001A1 (en) * | 2018-09-19 | 2020-03-26 | 语联网(武汉)信息技术有限公司 | Machine translation engine recommendation method and apparatus |
US11074420B2 (en) | 2019-06-25 | 2021-07-27 | International Business Machines Corporation | Self-learning input-driven brokering of language-translation engines |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1465018A (en) * | 2000-05-11 | 2003-12-31 | 南加利福尼亚大学 | Machine translation mothod |
CN1530861A (en) * | 2003-02-24 | 2004-09-22 | Language translating method and system | |
WO2010052696A1 (en) * | 2008-11-05 | 2010-05-14 | Ubstream Ltd. | Session initiation protocol in multi-cast/unicast system for connection between users behind nat and connection between server and users behind nat |
CN102253930A (en) * | 2010-05-18 | 2011-11-23 | 腾讯科技(深圳)有限公司 | Method and device for translating text |
CN102446169A (en) * | 2010-10-13 | 2012-05-09 | 张龙哺 | Translation system by utilizing online translation services |
CN103914446A (en) * | 2013-01-08 | 2014-07-09 | 崴强科技股份有限公司 | Translation method and system |
CN104011712A (en) * | 2011-06-24 | 2014-08-27 | 谷歌公司 | Evaluating query translations for cross-language query suggestion |
CN104090870A (en) * | 2014-06-26 | 2014-10-08 | 武汉传神信息技术有限公司 | Pushing method of online translation engines |
CN105051728A (en) * | 2013-01-03 | 2015-11-11 | 时新公司 | Data base query translation system |
-
2016
- 2016-04-29 CN CN201610272936.9A patent/CN106021239B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1465018A (en) * | 2000-05-11 | 2003-12-31 | 南加利福尼亚大学 | Machine translation mothod |
CN1530861A (en) * | 2003-02-24 | 2004-09-22 | Language translating method and system | |
WO2010052696A1 (en) * | 2008-11-05 | 2010-05-14 | Ubstream Ltd. | Session initiation protocol in multi-cast/unicast system for connection between users behind nat and connection between server and users behind nat |
CN102253930A (en) * | 2010-05-18 | 2011-11-23 | 腾讯科技(深圳)有限公司 | Method and device for translating text |
CN102446169A (en) * | 2010-10-13 | 2012-05-09 | 张龙哺 | Translation system by utilizing online translation services |
CN104011712A (en) * | 2011-06-24 | 2014-08-27 | 谷歌公司 | Evaluating query translations for cross-language query suggestion |
CN105051728A (en) * | 2013-01-03 | 2015-11-11 | 时新公司 | Data base query translation system |
CN103914446A (en) * | 2013-01-08 | 2014-07-09 | 崴强科技股份有限公司 | Translation method and system |
CN104090870A (en) * | 2014-06-26 | 2014-10-08 | 武汉传神信息技术有限公司 | Pushing method of online translation engines |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107894983A (en) * | 2017-11-08 | 2018-04-10 | 深圳市沃特沃德股份有限公司 | The switching method and device of engine |
CN107979856A (en) * | 2017-11-22 | 2018-05-01 | 深圳市沃特沃德股份有限公司 | Connect the method and apparatus of engine |
CN107979856B (en) * | 2017-11-22 | 2020-10-27 | 深圳市沃特沃德股份有限公司 | Method and device for connecting engines |
CN108319591A (en) * | 2018-02-05 | 2018-07-24 | 深圳市沃特沃德股份有限公司 | Realize the method, apparatus and speech translation apparatus of voiced translation |
CN109118109A (en) * | 2018-08-31 | 2019-01-01 | 传神语联网网络科技股份有限公司 | Quality evaluation based on ETM |
CN109118109B (en) * | 2018-08-31 | 2021-06-01 | 传神语联网网络科技股份有限公司 | ETM-based quality assessment |
CN109214011A (en) * | 2018-09-19 | 2019-01-15 | 深圳市合言信息科技有限公司 | It is a kind of by user feedback come the cognitive engine selection strategy of self-perfection |
WO2020057001A1 (en) * | 2018-09-19 | 2020-03-26 | 语联网(武汉)信息技术有限公司 | Machine translation engine recommendation method and apparatus |
US11074420B2 (en) | 2019-06-25 | 2021-07-27 | International Business Machines Corporation | Self-learning input-driven brokering of language-translation engines |
CN110532574A (en) * | 2019-08-20 | 2019-12-03 | 语联网(武汉)信息技术有限公司 | MT engine selection method and device |
CN110502762A (en) * | 2019-08-27 | 2019-11-26 | 北京金山数字娱乐科技有限公司 | A kind of transcription platform and its management method |
CN110502762B (en) * | 2019-08-27 | 2023-07-28 | 北京金山数字娱乐科技有限公司 | Translation platform and management method thereof |
Also Published As
Publication number | Publication date |
---|---|
CN106021239B (en) | 2018-10-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106021239A (en) | Method for real-time evaluation of translation quality | |
Bahia et al. | The welfare effects of mobile broadband internet: Evidence from Nigeria | |
KR102097426B1 (en) | Analysis method of fluidized population information capable of providing real-time fluidized population data by pcell algorithm | |
WO2020103677A1 (en) | Method and device for processing meteorological element data of numerical weather prediction | |
CN107451861B (en) | Method for identifying user internet access characteristics under big data | |
CN104572937B (en) | A kind of friend recommendation method under line based on indoor life range | |
CN102137155B (en) | Method for handling communication network quality complaints based on customer perception | |
CN105787574A (en) | Method and device used for providing seat information | |
Campisi et al. | Spatial variation in fertility across Europe: Patterns and determinants | |
CN103068035A (en) | Wireless network location method, device and system | |
Gu et al. | Modelling interprovincial migration in China from 1995 to 2015 based on an eigenvector spatial filtering negative binomial model | |
CN101257512A (en) | Inquiry answer matching method used for inquiry answer system as well as inquiry answer method and system | |
CN102711162A (en) | Method for monitoring network quality and optimizing user experience in mobile internet | |
CN107046691A (en) | A kind of LTE wireless networks quick site selection method | |
CN108205770A (en) | The acquisition methods and device of a kind of target medium | |
Visman et al. | Defining metrics for monitoring and evaluating the impact of co-production in climate services | |
CN104881412A (en) | Hotspot query method, apparatus and system | |
CN107347189A (en) | A kind of wireless private network base station planning methods of TD LTE 230 | |
García-Sánchez et al. | Decisive factors in the creation and execution of municipal action plans in the field of sustainable development in the European Union | |
CN107682864A (en) | A kind of base station construction method assessed based on coverage rate | |
CN114374953B (en) | APP use prediction method and system under multi-source feature conversion base station based on Hadoop and RAPIS | |
CN103188629B (en) | Flow bootstrap technique between a kind of networks with different systems and device | |
Lyytimäki et al. | Agenda 2030 in Finland: Key questions and indicators of sustainable development | |
CN114286408A (en) | Network performance optimization method, system, device and medium based on heaven-earth integration | |
Bosman | The changing position of society journals in Geography |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB03 | Change of inventor or designer information | ||
CB03 | Change of inventor or designer information |
Inventor after: Han Xin Inventor after: Zhang Chu Inventor after: Chen Gang Inventor before: Chen Gang |
|
GR01 | Patent grant | ||
GR01 | Patent grant |